Atomic Razor

Self-Sustaining Settlements

I just finished Thomas Thwaites's The Toaster Project, which is excellent. I reminded of that realisation I had school that every artificial thing in the world was made by tools that had been by more primitive tools. right back to the point where all you have is whatever you can find lying around in the environment. Also the later realisation that everything in the world that is made must also be sold. Thwaites necessarily cheats in making his toaster and he never actually toasts any bread with it. He was at the Royal College of Art, but talks to Jan Cilliers at Imperial. I have wondered whether you could build a car, say, from scratch at somewhere like Imperial. Obviously, you could certainly do a lot of stuff in a very well-founded university's laboratories and workshops. "From scratch" is doing a lot of work as Thwaites indicates, but I would imagine you could make plastic from crude oil in Chemistry or Chem Eng or Materials, perhaps Earth Science and Engineering. You could certainly refine metals. This might not be done so much these days, but there must be metallurgy left. So you probably make everything from scratch (from the raw materials) at Imperial, although, of course, you are heavily reliant on a lot of pre-existing equipment, facilities and tools. You could, of course, exclude more advanced technologies and shift things back a couple of tech levels, but not as far as the Paleolithic.

We bought David the Weinersmith's A City on Mars for Christmas because I knew he would like it. Self-sustaining settlements in space are hard for all sorts of reasons and might require magic tech (i.e. true Drexlerian molecular nanotechnology). Of course, you can do a lot on an Antarctic base or an oil rig or a submarine or an aircraft carrier, but you are still dependent on supply chains. You can ship a lot of workshop facilities to Mars. There's doubtless stuff you can do with the stuff lying around there. Obviously, you don't need to be self-sustaining anymore than an Antarctic base or remote city (Perth, say) is. You might want to be because you think the Earth's might get wiped out somehow, but you can just have a greater Solar System supply chain. And, of course, you don't have to worry about polluting space, although you do Mars or the Moon. For interstellar missions, you might want to find yourself a nice comet. Could you have one big enough to last an interstellar voyage of thousands of years (if you hop off to a passing star or an island hop via rogue planets)? You are going to loss some volatiles from your generation ship, but could you have enough of a buffer with your comet for a 1000 people for 10,000 years?

 

 

 

 

 

 

 

 

 

 

Twelve Months On...

Well, that's depressing. I thought I had posted something here earlier this year, but it seems that it was just over a year since my last entry. I have plenty of ideas for new posts in 2023, not that I can remember what any of them were now. There have been a lot of changes for me in 2023 with a lot more to come in 2024. I am hoping that some of those changes might involve doing more writing, but how am I kidding? Those books won't write themselves, but that was as true in 2006 or 1991. But I have  those two nearly written novels. Something can be done with them. Can't it? Can't it?

Actually, one thing did occur to me that I can recall. Consider all those imaginary friends of mine (I have a lot). We are soon going to arrive at epoch at which is going to be possible for me to finally get to read their books or, at the very least what the eleven tracks of Alan Fredericks's 1999 Mercury Award-winning album Again, from "Again" to "Emily" to "Stop" to "Again (again)" sound like. 

Yes, there's surely a novel in that.

In the Tavern

This is the picture that I inserted in my post yesterday at, I hope, a more reasonable size. The large one I sourced directly from Berfrois, "an online literary-intellectual magazine". So I acknowledge that source. This one is simply that image saved as a JPEG.  

I had not seen this picture until yesterday and yet I can readily acknowledge it as directly representative of my own life. I have felt and often do feel - if not as this very moment - exactly like I imagine the figure in Munch's painting feeling standing in the pub. I don't have a beard, but the man even looks a little like me. We have both experienced the Long Dark Teatime of the Soul. And for Munch and for Adams, teatime will never come again and there are no consolations to be contemplated. There still might be for me and even more for Justin Erik Halldór Smith, who, has I said is only 50 and is already never going to be in a position any more to look and feel like that in a bar, if we can just the way of out of the slough. "For most of 2021 I was indescribably depressed." And that was after he had given up drinking. (Yes, indeed.) '' “I think I’m a ghost,” I said. “I mean I literally think I’m the ghost of a person who used to live but no longer does.” [...] “I think I died in March, 2020,” I said. “My body was stored in one of the refrigerated trucks outside of Brooklyn General Hospital.” So things not exactly peachy there for Justin. That's more extreme than anything I have ever felt, in the past, much less now. I am not actually depressed, more anxious, stressed and with that sense, all too often, that something horrifically - existentially - awful is going to happen. And the question is how does one stop that becoming one's only mode of thinking as one moves ever closer to the precipice?

The Beginning of the End

• ChatGPT clearly is a big deal. We are nearer the Singularity now than we were a couple of weeks ago. ChatGPT is important because it is the first time that the We have got to play around with these kinds of Large Language Models. This isn't even GPT-4 and there are obvious ways of making ChatGPT better like giving it access to the FoAK and other sources canonical information sources. This is going to be big. Kurzweil's prediction of 2029 for the Turing Test to be passed seems very reasonable at the moment. A 2045 or 2038 Singularity can hardly be excluded. Think of this like Lynx in 1993 or, perhaps, better, Archie and Gopher in 1992. You get from there to Musk's Twitter in 30 years (also all the other stuff). Could Stephenson or Gibson have predicted it? Perhaps, but we wouldn't have believed it. So, who knows by 2052? I'll be 84. I am hoping that's the New 44. Also, natch, AlphaCode. It's not yet coding at superhuman levels, but as with go back in the mid-2010s, being able to code at the level of the average programmer is, in its way, solving programming.
• the National Ignition Facility reaching breakeven might be a big deal. We kind of know now anyway that with the magnets currently available an ITER-like system should work (presumably ITER with its outmoded magnets too, but it's clear there's a roadmap with better ones). Having different fusion paradigms is good. What would it like to work for a fusion startup? Plus, of course, AGI-like systems should help here. I wonder if inertial confinement fusion  might be better for Daedalus-type fusion rockets. Getting fusion rockets "soon" (20 or 30 years) would be a thing. It would be great to see people in Jupiter space in the 2050s. We probably don't need fusion to solve the energy crisis - we need to go from Fermi's pole in the University of Chicago squash to the commercial power reactor at Calder Hall in 15 years (13 for the submarine PWR on the Nautilus). That's 2035. Come on!
• Artemis 1 completed its mission, so no reason not to do a crewed test now. I guess you are still going to have to do an uncrewed lander test. I can't find my earlier post from about 2006 talking about the NASA Moon programme announced back then that I spotted earlier today. But it's taken a decade and a half or more to get here and, of course, we don't need Orion or SLS any ,ore. But still good to see this finally reaching fruition and very likely we are going to have a Moon landing 2025 or so. He-3 mines anyone. (James Nicoll dies a little very time someone suggests those, but come on...).
• David writing 1995 and All That, 17 years ago. Ten years is a long time or not depending how you look at it. I mean things feel very, very very different for me sat 54 than at 37. We were boys then, but didn't know. If we could go back... I think there is a phase change when you get to your mid-50s and realise there is nothing left to look forward to, but the grave. See this article by Justin Erik Halldór Smith. Edvard Munch, In the Tavern, 1890 via https://www.berfrois.com/2022/12/justin-e-h-smith-hell-yes-life-is-good/?utm_source=substack&utm_medium=email. That's pretty much how I feel and look all the time these days. But he's only 50 (and has already given up the sauce) and gets to research and teach cool things (Russian cosmism!) and has already had an asteroid name after him (how did he wangle that?). "I am co-author, with D. Graham Burnett and Catherine Hansen, of a work of historiographical metafiction, appearing in 2021 from Strange Attractor Press." What? There's always more. How had I not heard of him before? I mean that's the thing. Why not I get to personally hang out with these people? Would I be way of the pace? How can I get to meet them? Yes, this is essentially the theme of Little, Big. Indeed. There's something there. Smith mentions Kieran Setiya's Midlife: a Philosophical Guide, which I should get. Then again is there any point when you are living through it. Is it going to tell me anything helpful. And, more to the point, how come all the people interviewed in the classic studies of midlife crises are in the 40s or younger? They are children. Why is no writing about us? What am I supposed to do or feel at 54 when I am staring into the void and it seems terribly, terribly close in a way it didn't even a year ago? Humh... We shall see.

We Need to Talk About Uwe Johnson

I am one of the few people on the planet to have read Uwe Johnson's monumental Anniversaries, (at least I think I have read it all; it's possible I someone managed to skip one, presumably short section) whether in the recent complete English translation or in the original German. But is it any good? Should you put aside a year (because that's the natural length of time it will take to read to it) to plough your way through it?

Johnson sprang to fame in 1962 when he won the the Prix International for Mutmassungen über Jakob, "Speculations about Jakob". (Jakob is a character in Anniversaries, although this is no Tristam Shandy and we do get in 1600 pages the consolation of learning the exact details of his begetting of Marie, one of the two major characters in the novel , the other being her mother, Gesine.). The significance of that, of course, is that in the previous year the prize was famously awarded jointly awarded to Samuel Beckett for "The Trilogy" and to Jorge Luis Borges for Ficciones. Beckett was already well-know but the publicity owing to the award to Borges quickly led to the formation of an intense global cult around the writer and his works that very much continues to this day. So Johnson, at just 28, was following in some pretty august footsteps. Nabokov complained that he was disappointed that all there ultimately was to Borges was a few short stories and though Beckett wrote several novels, he is best known for short works. Not so Johnson. Anniversaries is colossal. A section a day every day for a year (that's a leap year, 1968) plus a prelude section and an appendix to the second book of the tetralogy containing some reminiscences of one of the characters. The whole thing weighs in at nearly 1700 pages. But is it actually any good?

The translation is workman-like enough. There's the occasional nod, but that's inevitable in such a lengthy work and there are certain stylistic decision I'd have called differently Station Street in Jerichow, for instance, should, surely, be Bahnhofstrasse. Now, Jerichow is a fictional place and I suppose it's possible that the street names in its real model were in Plattdeutsch and Johnson translated them into standard (High) German. Is there any internal (or external) evidence for that? Certainly, Luftwaffe could be left as Luftwaffe with fear of confusion as could the names of newspapers. All of the units could have been left as they are in the original text even if that might discombobulate people unable to think at all in kilometres or kilograms. (There is a German inch, but I think it might be exactly 2.5 cm and a French pound, la livre, 500 grams, but I don't know if they have that in Germany. "Aujourd'hui, le terme Pfund est toujours d'usage courant et réfère de manière universelle à la livre de 500 grammes." If you say so, French FoAK. Also, of course, how much is the term actually used in practice today or in the 1930s-1980s.) But, of course, you are always going to lose a great deal in a translation. Johnson might be a master of German prose, but one is never going to know that.

And thereby, I think, lies the major problem with the book. If you compare to other big, "hard" books in English that one might reasonably compare it to, Tristam Shandy or Moby Dick or Ulysses or Gravity's Rainbow (Pynchon was born in 1937) or Infinite Jest, one can't read one's way into the book in the same way because there's always the gap of the language there. But, of course, it's not just the language. Johnson was born in 1934. He died in 1984. (So, unlike both Beckett and Borges, he didn't make into his late 80s.) He lived in New York for several years in the 1960s, where much of Anniversaries is set and, although the book is certainly in some ways a New York novel, it's really and increasingly a Germany novel. Part of this can be ascribed to the long gestation period of the fourth book of the novel (it's definitely a single novel and not really a tetralogy except largely in terms of publishing schedule and convenience). Johnson left New York and went to live in Sheerness on the Isle of Sheppey, where, doubtless, the eternal, grey vistas of the Thames estuary from the upper stories of his house reminded him of his beloved boyhood Baltic. Johnson liked reading newspapers and many sections of Anniversaries begin with extended quotes from that day's New York Times. But, in provincial Kent, 1967-8 copies of the New York Times were hard to come by. We get a lot more of Gesine's life in the immediate post-war Soviet Zone and a lot less of New York. It's perhaps no surprise that the fourth volume took more than a decade to appear after the third. But here's the thing. Johnson might have the last decade of his life in England and died there (there are even early passages of Anniversaries set, perhaps improbably, in Richmond where Gesine's father works as a carpenter in the early 1930s; does Patrick Wright point to a family connection of Johnson's for that, I wonder, in his book on him?) and Gesine lives in New York (in the same apartment as Johnson did in our reality), but it's a very German novel. I have no idea what to compare it to. Johnson is, despite his American and British connections, a very middle-century, Mitteleuropa writer. Who is like? Heinrich Böll? Günter Grass? W.G. Sebald? What classic German writers is he channelling? Thomas Mann in The Magic Mountain? I know almost nothing about German literature. Johnson was drenched in it. And German culture. Gesine isn't the exact same age as Johnson (and she's a woman) and her family history is definitely different, but obviously aspect of her life match  Johnson's. But also, I suspect, isn't very possible that a lot of the things that happen to Gesine at high school and the people she knows are based on stories that Johnson knew from his life in East Germany, from other people he met or things he read. People who lived in East Germany (or even perhaps in West too) might smile wryly in recognition of the things that go in post-war Jerichow and Gneez.

Now, we might say that Johnson liked hanging out in the pub in Sheerness. He must have been a fun guy, at least at some level. But Anniversaries isn't really a fun or engaging book. It is not like Gravity's Rainbow, a book one says to oneself on reading, why didn't I read this book decades ago? Nothing really happens. OK, there's a vague sort of kidnap section at one point early on and perhaps there's some deep significance to that that perhaps fits in with what happens in Finland right at the end of the book, but if there is it's lost on me. Certainly, one learns a great deal about everyday life in rural eastern Germany from the early 1930s to the early 1950s (and, yes, in New York in 1967-8), but it is not generally very thrilling, viscerally or psychologically. I suspect I am missing an awful lot, probably if you are a Germanophone it's full of guffaws and the pleasure of recognition. Gesine never makes it to Sheerness, although she presumably does to Czechoslovakia. Well, indeed. Johnson was clearly very exercised by the Prague Spring, which in the early 1970s was certainly very salient. Gesine is, to my view, a very passive and cypher-like character (but I am not sure she is supposed to be). She works in a bank and is pressurised into taking on a special assignment to Czechoslovakia for the sinister vice-president (and éminence grise) of the bank  This again makes it sound more exciting than tr actually is, especially as it drawn out over a thousand pages or more. The novel (spoilers, I guess) ends with Gesine and her daughter, Marie, about to arrive in Czechoslovakia on... 21 August 1968. The day of the Warsaw Pact invasion of Czechoslovakia. Again, this is obviously supposed to be significant. Her mission is now going to come to naught as the Czechoslovakians aren't going to be in the market for Western loans for a while anymore (I seem to recall this is discussed in various of the NYT extracts so it was obviously a thing at the time). Is the implication that Gesine is going to die in Czechoslovakia? I don't know. I don't think so. Why should she? Sure, it's a bad time to arrive, but she's a US citizen by this point, I am pretty sure, so unless there's something I am missing she will eventually get out. Perhaps Patrick Wright's book gives some clues. But that's apparently 700 pages long, which, I suppose, appropriate. I do have a copy.  

There's the occasional little post-modern gesture, but it's not really that kind of book. It is, as I have suggested, a book for people from the middle of Europe in the middle of the twentieth century. It's 2022 now and I am don't have the knowledge of and emotional investment in the milieu that Johnson had an a German intellectual of the late C20th. Beckett and, for me, Borges may abide, but though I can say I am glad I read it, it will give me bragging rights in certain circles, I can't say I enjoyed it in the way I enjoyed other big books. But I am open to persuasion as to what I might have missed. 

 

Great Minds...

I really ought to blog more. Partly because I like it and partly because I like reading my old blog posts: wow! That guy must have been really something! Anyway, a couple of months or so ago, I read something about the acceleration of the expansion of the universe starting about 6 billion years ago and thought to myself, heck, I mean come on, if that's not evidence of the aliens messing around with the physical constants, or at least gravitics, I don't know what is. (I mentioned to my SDS chums.) Of course, I'd already had the same thought myself in 2016: Dark Energy. So, great minds do think alike and, presumably, many others have already had the same idea. But it's something for that space opera novel if no-one's already been there.

Plancktech, Platotech and Spinozatech

Yesterday, I was looking at some tweets from the esteemable Winchell Chung that showed various maps of the galaxy. These included inner and outer galactic habitable zone boundaries. WEndell commented that the radiation environment towards the centre of the galaxy is pretty intense. This got me thinking about Vinge's Zones, which I had thought were artefacts imposed by the Powers in the galactic halo. So you can't have an electronic or biological computer at the centre of the galaxy, but perhaps you could have a mechanical one. There's a novel in that. Also I'd never come across an outer habitable zone boundary, but I recall a paper in which someone suggested the Jupiter brains would migrate to the edge of the galaxy because it's cooler and thus the thermodynamic efficiency of your computronium is higher. This could be a bit like Vinge's Powers. Of course, in A Deepness in the Sky, we have a planet out of place that's there's something weird about, so I tweeted

I forget now if much is ever hinted at about that by Vinge. Femtotech? Plancktech? Platotech? Will be there some big reveal in an as yet unpublished novel?

Plancktech, the technology of manipulating objects at the Planck scale, just popped out once I'd written femtotech and the word is used in a Reddit post. It then occurred to me that, of course, you also need a technology to manipulate objects in the realm of Platonic ideals. This is obviously Platotech. I can't find an earlier use of the word, but the idea is certainly not original to me. Anathem is an obvious example, there must be others, and it does appear, unnamed, in both my works in progress (the dating novel and the steampunk one).. But then I thought we need a more general term for technology for manipulating objects in all classes of ontologically possible universes. But what to call that? Surely someone must have discussed the idea of the whole complex of all possible universes? Tegmark does discuss at least something like this, but we must have got it from somewhere and there must be earlier C20th or even C19th philosophers who explicitly talked of this. John Leslie sprung to mind and after a little poking around I hit paydirt and was able to emit this tweet

John Leslie's webpage mentions "a Spinozistic pantheism of infinitely many divine minds, each of infinite complexity; the structure of our universe, and of countless others, would be part of that complexity", so Spinozatech it seems it is!

Which means that we are back to the C17th and Spinoza. I even recalled Spinoza's version of the Ontological Argument from reading Ethics back in 1991, which is essentially that God must exist because in an infinite number of universes some must contain God. So that's perfect.Spinozatech and Platotech are recorded anywhere as far as I can see at the moment, so I am claiming credit for these terms until proven otherwise. And not a bad night's work!

A Recursion in Metastories

Well, the Razor was here on 31 December 2020. I was surprised and disappointed to see that I didn't post at all in 2020, that the last post was on 31 December 2019 with Twenty for the 20s. Well, he have had our first pandemic. Was it that day that the Chinese authorities informed WHO that there was a new and virulent coronavirus in circulation in Wuhan? Has Shanghai really been back to normal for months? And the death warrant on tens of thousands of people in Britain has already been signed. I just hope that I am not one of them. COVID-19 is going kill tens of millions of people globally. That's a Level 7 Black Swan, so probably comparable to the First World War or the Great Depression in terms of its impacts. Of course, there could be worse to come.  

Unlike some - a lot of - people, my life didn't really change in 2020. I work from home, on West Coast time. No new swathes if time were suddenly opened up for me. Yet that is no excuse for not having blogged. I did a tiny bit of writing (because of Zooms with the London Literary Lions) and read 31 books, but quite a few of them were Black Archive Doctor Who monographs. 31 doesn't even begin to make an impact on the number of physical books I bought much less the virtual ones. Looking back over the Razor, it it is full of exhortations to write more, blog more, read more, do more. See, for instance, this entry, not even a New Year resolutions one from 2015. Of course, this is not surprising. The things I like, the things I am interested in, the things I do or want to do, the type of person I am, is much the same now in 2021 as it was in 2015 or 2004, 17 years ago when David and I started the blog, or 17 years before that in 1987. I know what I want to do and I am still mostly not (or, to a certain, not not) doing. You might say just admit it, Paul, just give up, you are never going to do it now, if you were, you'd have done it two or three decades ago. But obviously I can't give up. Not doing it won't make me happy and giving up pretending I am going to do it won't make me happy. There is No Alternative.

I got Uwe Johnson's Anniversaries. This is a very long (nearly 1700-page) novel. Normally, that would be a hopeless project, although I have ploughed through Infinite Jest and Gravity's Rainbow and the whole of The Baroque Cycle. But it is written in the form of one section a day for a year, although starting in August 1967, so I have started reading it one section a day and will finish it on Boxing Day. I also have the 25th anniversary edition of Brian Eno's A Year with Swollen Appendices. This does began on 1 January (1995). Even with the appendices, it's much shorter than the Johnson, so might be impossible not to read at a faster rate. In his new introduction, Eno provides a long list of concepts, provided by the usual suspects including Kevin Kelley, Danny Hills and Stewart Brand, of concepts that either didn't exist at all or have changed their significance greatly or have become much better known than they were in the mid-1990s. 1995 is, of course, a turning point with the Netscape IPO on 9 August changing the world forever. One imagines that passed even Eno by. No mention on 9 August, but quite possible Netscape is mentioned elsewhere in the book. Eno exchanged ~100,000 words of emails that year with Stewart Brand. But then Arthur C. Clarke had a book of emails published in 1984(!). But, yes, it's easy to think back to 1995 and forget just how different a lot of things were then. It was at the end of the C20th paradigm of telephones and motor cars that runs from perhaps the 1920s until about 2000, although the new paradigm is not full established until ubiquitous smartphones about 2010. But certainly, in many ways, 1995, has more in common with 1969 than 2021. Everything always stretches back further than you think (the past is both more and less like the present that we realise). It's still very much an analogue world (no Google, Wikipedia, Facebook, Twitter or web APIs). It's why I think we are living in hypermodernity. Under postmodernity, we could imagine the World Brain, under hypermodernity, we could instantiate it. 

Of course, 17 years from is January 2038. That's frighteningly not very far off. And who know's what might happen on before 03:14:07 UTC on 19 January 2038. I just hope we are all around to see.    

Twenty for the 20s

Back in 2014-5, after I had been to Arvon, Simon Ings suggested a write  fix-up, which was a very good idea, and I did a series of posts of the world in 2034, 2054... which, I think still stand-up pretty well. I am currently, when things like Christmas, I know, don't get in the way doing in the Institute for the Future Coursera course on futures thinking. Their focus is typically on the ten-year horizon, so here are my Twenty for the 20s:

1. Level 10 Black Swan. Which would be an extinction event. It's quite hard to wipe out the whole human race, but not impossible. No Personal Nanofactories by 2029, so I think this won't happen.
2. Level 9 Black Swan. The collapse of the current technological civilisation. Again, quite hard to do, but again certainly not impossible. Full-scale thermonuclear exchange? Natural or artificial pandemic? Climate breakdown-induced breakdown of global food supply chains? I don't think this will happen, but, of course, that is in large part because I don't want it to.
3. Level 8 Black Swan. That WWI-level event that we have felt lurking our there. Given the current nature if the catastrophe, this is quite likely. You simply can't have the kinds of people in charge who we have in charge in certain countries without something catastrophically awful happening. I think there is a real chance of one of these. History might well be about to return to Terra and change all our lives forever.
4. Heterodox economics. There have been hints towards this including with the Labour Party, but Corbyn was never the man to sell cybereconomics and Beerism. Might Cummings be? Not sure yet about Modern Monetary Theory, but there are enormous possibilities outside the neoliberal consensus. 
5. Neonationalism. We will see how that shows in the UK over the next four years. Can only hope that we can get our acts together to avoid the worst consequences of this. But I am not optimistic. There is that Level 8 Black Swan lying in ambush.
6. Scottish Independence. Both more compelling and harder to achieve than in 2014. That hard border along the Cheviots is really going to be an issue. I think they missed the Scots chance, which is both bad and good.
7. Globalism. The African decade? We have already seen breakouts by companies like Samsung and Huawei. But what about a Nigerian or Brazilian company?
8. A(G)I. We have had an AI Summer. Everyone is expecting an AI Winter. Clearly, Deep (Reinforcement) Learning is not the whole answer. Nevertheless, how many of us expected Go to be solved and solved in the way it was? My hope is that if we let a thousand philosophy bloom, there is a chance that some PhD student in China can come with an idea that a postdoc in Canada can leverage and suddenly open up a whole new front. Plus, we are getting the fourth edition of Russell and Norvig early in 2020. I suspect some time before the end of the decade we will have systems that meet the definition of domain-specific AGIs idiot savants/dog/cat/great ape-level AGI. And remember, humans didn't even evolve language for something like the first three-quarters of existence, which is incredibly freaky, and suggests you can glom language on top of non-linguistic GI.   
9. Self-driving cars. A hard problem, but think where we were in 2009. Does not require full AGI. I hope to own one by 2029.
10. Electric cars. These are already a thing in that I see at least one pretty every day just driving around Preston. I even saw an electric scooter outside a kebab shop the other day. Should be fairly ubiquitous by 2029.
11. Hydrogen/methane economy. If we want to completely decarbonise the global economy by 2045, we are going to need a lot of local hydrogen and methane (for instance, from biodigesters for use in ships) production, so I hope to see the infrastructure for this becoming common.
12. Fusion. There are quite a few fusion startups around. Can one make the breakthrough? This is the thing we need to solve the energy crisis fully and also to open up the Solar System.
13. Quantum computing. This won't break encryption because we already know of types of encryption that QC can't break. It will certainly be good for simulation of quantum systems and quite possibly for various database-related applications and for brute forcing certain problems. Whether we will have desktop quantum processors by 2029, even in the lab, is another matter.
14. True Drexlerian Molecular Nanotechnology. We haven't seen as much progress in this area as we expected in the 1990s. We have seen an AI Summer and we are overdue an NMT one. It is not as though it has suddenly become a bad idea. Might well be helped by quantum computers, which will be good for simulation of chemical systems.
15. Return to the Moon. We should get this. And a moonbase. Yeh! But will it be Moonbase Britannica with the bally Union Jack on it?
16. Heavy Launch Vehicles. If we get the Starship and the New Armstrong , things could get quite exciting quite quickly. Orbital tourism? asteroid mining? Mars base? A golden age of uncrewed missions to the outer planets?
17. Life extension. Would be good to have solid progress here. I do think it is going to be an definite theme of the 2020s (certainly for me personally!), A Methuselah Mouse would be good.
18. New Physics. I fear we will be arguing about string theory and its lack of any empirical backing. Would be fun if the X17 particle is real, but I'd wager it isn't. What would be great would be to have a AGI come up with a Theory of Everything. (I do expect at the least that an AGI will have achieved a first in physics from Oxford by 2029.) What about Dark Matter? Or Dark Energy? Or Inflation? Physics feels as though it has run into the sand right now. Might we be in a Physics Summer by 2029?
19. SETI. We are certainly going to know a lot more about extrasolar planets? Might we have found life on Mars? What about those disappearing planets? Fast radio bursts? Perhaps we might observe something for which the most plausible explanation is a large-scale engineering work in the universe. Or even get the Message.
20. Technological Singularity. Not, I think, by the end of 2029. But I do think we will start getting some crazy stuff after 2025. I think it might feel closer by then. But I am still expecting the Singularity at 03:14:07 UTC on 19 January 2038.

Well, the Razor was here on 31 December 2009 and I can only hope we are still here on 31 December 2029. We will see.

You Can't Get Wet in a Simulated Rain Shower

Someone in of the articles in Dennett's and Hofstadter's The Mind's I propounds the notion that though you can simulate weather in a computer you can't get wet in a simulated rain shower. There is thus a difference between the simulation of something and the the thing itself. And so a simulated mind is not a mind.

The argument does have a certain rhetorical force, but there did seem something slightly fishy about it. Dreyfus may have argued in the 1960s that a computer would never play a good game and Deep Blue does not play chess anything like a person does, but it is hard to argue that AlphaZero doesn't play a good game of go or chess or shogi and it does it in a more human-like way. Similarly, a self-driving car is driving itself and a translation program is producing a translation in exactly the same way that a translator produces a translation. OK, not in exactly the same way, the translator can feel things and want to do things in a way the program can't (yet), but the translation is a translation.

The thing is that a simulation of a mind is a mind. It might not be a mind like ours, but it is still a mind. The simulation of the mind and the mind are doing the same thing and thus they are not like the simulation of the rain shower. The output of the mind and the simulation of the mind are the same thing. Which is not the case with the real versus simulated rain shower. 

I recently started listening to the Lex Fridman AI podcast and one of the early episodes is a conversation with Christof Koch, president and chief scientist of the Allen Institute for Brain Science in Seattle. Koch comes out with exactly the rain shower argument so obviously whichever paper it was in the Mind's I made a big impression on him too. I see he has a book due soon The Feeling of Life Itself – Why Consciousness is Everywhere But Can’t be Computed, Clearly, he has a counter argument to mine. It probably doesn't feel like anything to be AlphaGo or Shannon's mouse or a self-driving car. It probably does feel like something to be a chimpanzee or a dog or a cat or bat or a mouse or a New Caledonian crow or an octopus. Does it feel like something to be a bee? Possibly? What about a beehive? Or a forest? But the rain shower argument is clearly fallacious. Koch does have a counterargument to my argument. But it would be helpful if he had some Reader's Digest version of his counterargument rather than putting forward an obvious strawman. 

Some suggestions for those arguments might be

• the mind is very complicated
• the mind is very complex
• the mind is an analogue computer
• the mind is an oracle
• the mind is a quantum computer
• Searle's Chinese Room
• Penrose-Lucas argument
• physical embodiment
• social embodiment
• the soul.

There are counter-arguments to all of these, but at least we shouldn't be expected to take seriously an argument that might have emotional power, but which doesn't stand up to a cursory analysis. 

Mapping Into Abstraction Such a Territory

The other week, S was reading one of the Stuart MacBride Oldcastle novels. Oldcastle is a fictional city in Scotland. The book she was reading had an OS-type street map of Oldcastle. So where the heck did that come from? This isn't just a map on a napkin or even the kind of thing you might find in an RPG supplement. (There are many wonderful things about White Dwarf 42 and one of them is that inckudes the first part of Irillian.) Someone has to put quite a lot of effort in with the mapping software. Which would be easier if tey had taken an existing city map and just flipped it somehow and then renamed the streets, buildings and districts. This is what I suspect was done as it would be the easiest thing to do, but I don't which city map they have used as the basis. If Oldcastle is "supposed" to be Dundee, you can't just flip the Dundee map as Oldcastle, like Dundee, is on the north of the estuary. Newcastle might be a possibility, of course.

The Old Devils is a very topographical novel. But is it just the topography of Swansea and the Gower with which Amis was very familiar or did he make it up?

When I was a child, one of my concerns with both plans of schools and maps of towns was missing something out. This couldn't possibly be a real school or a real town because a real school or a real school of its type would have these certain features that I have neglected. Getting hold of plans of schools in the early 1980s was not easy, but maps of towns were a bit easier. One could study enough maps of towns and produce a plausible enough looking map of fictional town I should think and perhaps that is what MacBride or someone in the design department at his publisher did.

But we know have generative adversarial networks. And, yes, at least one group has used GANs to produce images of imaginary cities. What is happening here is generating satellite images of Venice in the style of Los Angeles and vice versa and then using the filters to generate the imagesd based on handdrawn maps. You could you could then produce a map based on the satellite image. This doesn't help with missing out the police station.

There are hundreds of towns and cities in the UK. One could imagine using a GAN to generate imaginary maps that are difficuklt to distinguiish from thre real thing. The simplest system would just look at the road layout (plus indeed, infrastructure like railways, canals and rivers), but more sophiticated ones could look at the buildings and even the names of the streets. Yes, you might want some hand annotation, but I suspect you could generate reasonably convincing maps.

Of course, getting access to the maps might be no easy thing. Yes, you could have a go with Google Maps, but really you would want pretty full GIS data, so you are not using an image of the town. And some of this might require something closer to the solution of the general AI problem. How does the system know how many primary and secondary schools there should be or where the teritary college should be or the hospitals. How much knowledge of the world does it need and how much can it do just on the basis of features extracted from maps? Does it help know that a town the size of Preston will have a large hospital or can you just learn that with knowing anything about hospitals.

It's like the old Indonesia problem. Indonesia is a country in South East Asia. Malaysia is also a country in South East Asia. Singapore is a country in South East Asia, but it also a city. So Indonesia is like Malaysia is like Singapore. But Singapore is like Jakarta is like Kuala Lumpar. As countries, Indonesia and Malaysia are a lot more like one another than they are like a city state such as Singapore. But how do I learn this? As a person I have a concept of South East Asia, countries, types of countries, cities, capital cities, modern megacities in Rising Asia. But where do I get these categories from? How do I build a system that can learn these categories without having to be told them explicitly. There are knowledge representation systems such as a Cyc or Wikipedia, but what's the the state of the art in knowledge engineering. I live in hope that we will see a Coursera course on this one day.

It's not just fictional city maps that GANs could help with. There are also fictional languages. I will talk more about imaginary languages later, but GANs might be a great way to generate the kind of plausible Indo-European language I dreamt of as a child.

Even the Longest March...

Following on from John's comment, I have now recovered the latest draft of the novel from Gmail, so I now have a monthly archive for TM on this computer for the first time since February 2015 after which I would have been using the Windows machine. Looks like it was February 2017 since I was doing any work on the novel, which is depressing, but makes sense. I did the Pandemonium of the Sun work in Seattle and though it is July 2017 since I got back, just blogging since then. Or if there is anything after that it went down with the Mac, but there couldn't be much. The point is to do what can be done. There is a lot of work to be done. I have a 323-page manuscript. Much of that useable after revision, indeed, of course, much of it has been revised already serveral times. Some of it will, sadly, need to be discarded. Some new briding material will need to be written. Yes, I would like to go on a retreat to Arvon, such as a Clockhouse retreat at The Hurst. But as ever the issue is not retreats or courses, but focus. Just doing the damn thing and learning from it so that next time can be better.

So taking one step at a time...  

The Outer Zones

When I first started at my previous job ten years ago in April, I line-managed a team of three people. One was an contracts executive. I didn't really get on with her. She read Daily Mail and had the typical price of everything and value of nothing attitude of the type and was incredibly fussy about what she woukd eat, which made planning team lunches frustratingly difficult. She had a higher opinion of herself than I did of her, my opinion of her not helped by the fact that she probably earned more money than I did. But she once make a good point. Which is that Zone 2 is huge. She lived in, at the time (she latter moved to Sevenoaks after she got married) in somewhere like Blackheath. Now, Blackheath station is, like Kew Bridge actually in Zone 3. Lewisham is on the Zone 2/3 boundary like Turnham Green. From I can gather Blackheath is a fair-sized area, so depending which part exactly she lived she might have lived in Zone 2.

But her point stands whichever way. Stamford Brook is definitely in Zone 2 and I had a girlfriend, R, who lived in a block of flats opposite. But Stamford Brook feels like quite a long way out. There are another 4 zones to go at leat and there are the special zones towards Amersham. Brentford is Zone 4, but seems to have far more to do with Stamford Brook than Stamford Brook has with central London, alkthough if you lived in Stamford Brook, you might not have much reason to go Brentford. 

I am always frustrated by London guide books of one lind or another that barely if at all get outside Zone 1. Zone 2 is big and a lot of people live there and there are lots more zones and lots more people beyond it! Of course, they are just boring suburbs, but they are boring suburbs that are still full of things with national and even global resonance in a way that isn't true of the suburbs of Preston. I like suburbs because a suburb s where I come from (and where I live), but slso where I want to escape from. As an example, consider Hugh Ronalds. I had not hesrd of him until just now, just he was a renown nursuryman who was born, lived and died in the same house in Brentford, the many of the contents of which eventually ended up in a museum in London, Ontario! Now, I wonder if Robert Rankin picked up that. Iain Sinclair, natch, has done stuff about the outer zones too, so one can flan in them. One could, indeed, as I have discussed flan in Preston. (As John would say, I ought to flan more here.)

When you drive out of London on the A316, it is miles after you get to the roundabout in Twickenham before you get to the M3. That's a huge area of suburbs in Zone 4, 5 and 6 in Richmod-Upon-Thames and Hounslow boroughs. There's even a Vicarage Road with a Beechwood Avenue off it near the Sunbury interchange This might be Shepperton - J.G. Ballard country. What it would have been like to have lived - to live - in that road? What manner of things have gone and do go there? These are the fractal city. There is always more.

500 Words

That's what a Naomi Alderman told me at Arvon in the summer of 2015. I can do that, she said. And she was right, which was howm, under the current dispensation too, though I was in London, I managed to get the first draft of TM finished. But the problem is that TM needs a lot of work. And very little of that work will be fun. I don't know how you could even estimate how long it would take to revise TM. At 500 words a day that's 200 days, 40 weeks. And, in theory that's a maximum because because you should be able to revise more quickly than write first draft because you can keep at least some of the words!

So David right too. I should be able to write 500 words a day. There's no excuse, no reason to wait for circumstances to change as that change might never come. And there's LifePlan2025. I know, I know. But what do you do? I haven't that much searoom now. I'm mearly 51 already. I have been staying this for years now, decades. The time to start is long past/passed.

But then what to write? My 2000s or even 1990s ideas, Fear of Thirty, Eleven Short Stories About Love, British Grove, just don't have the resonance given where I am now in terms of life stages. I can't face Fear of Fifty and I'm probably at least three years to lare for that one anyway. I don't feel ready for The Visibilities yet. Pandemonium of the Sun is a Seattle novel and woukld require quite a lot of research into at least early 1960s Seattle if not Pynchon himself such as getting the ship he served on correct. OK, I know a minor detail, but the downside of historical fiction is it can be difficult to take the past for granted and get on with the story, although I did manage that with The Mundanomicon. In fact an argument against revising TM is that I ought to read a lot more early C20th literature as well as books on the period to get back into zone. I had an idea for a novel about five characters, all of the immigrants in one sense or another in Seattle, perhaps, or even Preston. So, you would have the scion of the wealthy family who has lived a completely globalised life, the scholarship student to a Western university who is now a civil engineer, the civl engineer whose credentials are worthless in the West and so is a taxi driver, the person from a remote village who managed to make it legally to the West and works in a fast food restaurant and the illegal immigrant who has got into the country and will do whatever necessary to survive. But this would require years of research. 

Simon Ings suggested writing short stories and fixing them up. That has a definite appeal. Steve said it is better to write short stories as a way of improving as a writer. If one could write 500 words a day, one could write an revise a short story a month. After a year, that would be twelve short stories and one might get one published possibly. You might recall I did entries on the world in 2034 through 2114 in 20-year steps, which shows how long ago that all was now (the end, admittedly, of 2014). An Accelerando-style novel is appealling. You could have 11 10,000-word novelettes going from 2024 to 2124. But, of course, it is coming up with the plots for the short stories not just the chrome that goes around them. The migrant stories could be part of this.

But, you see, I just did it there. 632 words in fact.

The Name of the Doctor

My (strictly speaking my eldest brother's) copy of The Making of Doctor Who is in storage so I can't look up what it says about the Doctor's name, but, as I recall, it states it is a mathematical formulae and, indeed, gives it as a string of pseudo-mathematical symbols. This might be where the Bristol Boys got Theta Sigma from. As Miles/Wood point out, it is likely that Baker and Martin thought they were really revealing the Doctor's true name. The symbols in TMoDW might be where the notion of what Od High Gallifreyan characters might look like as seen in "The Five Doctors" and elsewhere comes from. 

So we can never be told the Doctor's name because it is an equation or some kind of technical data. This might fit with Miles's concern with the value of Time Lord (especially the Doctor's) biodata in Alien Bodies. If Time Lords are loomed, then their name could be their uniqie biodata, just as our own DNA is unique (but, natch, identical twins). The biodata is far more than mere DNA according to Miles and that of senior Time Lords especially (just think of all that wibbily-wobbily and the nature of the Time Lords as a race intimately connected to time). The Doctor has been Lord President and might have been the Other and well as being the hero of the whole series.

We can't be told the Doctor's name because it is unpronounceable. Knowing something's true name give you power over it and a Time Lord, especially the Doctor, is an extremely powerful thing to have power over. What they though do we have in TMoDW. What does River tell the Doctor in "Silence in the Library"/"Forest of the Dead"? What does Clara read in the book in the TARDIS library in "Journey to the Centre of the TARDIS"? It might use a link to the name or the beginning or other part of it.

But it not just the Doctor amongst Time Lords who has what appears to be an alias. The first Time Lord we meet is Susan, which is a standard English language name. But what is Susan's real name? Does it have the same meaning in OHG as Susan in Hebrew? Lily, so that could be a lily-like Gallifreyan flower. Or does it just sound (a bit) like Susan? Or was it just plucked from the air as sounding plausible for a 15-year old in 1963 London (7th most popular girl's name in the UK in 1944)?

The second Time Lord we encounter is the (Meddling) Monk. Then we have the War Chief, the Master, Omega (created by Baker/Martin, so compare Theta Sigma) the Rani, the Valeyard and the Corsair plus probablly some others I have forgotten. But then of Time Lords who do have "real" names. We have Romana (short for Romanadvoratrelundar, of course), Borusa, Drax, Rassilon and many others. The Doctor Who Wiki category for Individual Time Lords has 574 entries.

We are never going to be told the Master's real name anymore than we are going to be told the the Doctor's. But is Romana's real name anymore Romanadvoratrelundar or Professor Chronotis's Salyavin than the Doctor's is the Doctor? Professor Chronotis is a particulary interesting example as "Professor" is a title and "Chronotis" a vaguely classically sounding play on time. Is "Chronotis" like "Who"? But it is noteworthy that although deuterocanonical sources and the programme's titles for many years as well as popular culture make free play with "Doctor Who", the Doctor themselves is never referred to as "Doctor Who" expect by WOTAN in "The War Machines" and the joke. We have Doktor von Wer  in "The Highlanders". The "von" turned out to be exactly right, the Doctor is a lord. And we can allow him his little jest. There a question that must never be answered.

"There is only one war that matters"

I am still around. I was at S's parents last weekend and have been doing the first part of the Coursera course on TensorFlow. I have also finished my (rapid) great watch of Game of Thrones in time for the final season starting in April. This is, naturally, something about which I will have a great deal more to say. The short version (or rather part of it): although we haven't yet, of course, seen Season 8, we already know, after Season 7, that we are going to be disappointed. GoT is like Mad Men: tosh, superior tosh, but still tosh. Like MM, I will be absolutely bereft when it has finished for good.

OK, a bonus question to get us started: how Daenerys learn that Jon & Co. are stranded on the ice lake surrounded by ice wraith? How long does it take Gendry to get to Castle Black (or is it Eastwatch-by-the Sea?) Dragonstone must be at least a thousand miles from the Wall, probably more. And does Dragonstone even have a working ravenry? And then how fast can dragons fly? And how does Daenerys know where they are?

"This Year You Write Your Novel"

Yesterday I read a review in the NYR of the second volume of the improbably named Zachary Leader's biography of Saul Bellow. Leader, of course, is a big mate, of Marin Amis and wrote the authorised biographyt of Kingsley, which I have read. Martin is a big Bellow fan and I think, somewhere, describes him as one of his literary fathers along with Nabokov and Kingsley himself. I haven't read any Bellow, although I have at least one of his books somewhere in my storage unit. "John Podhoretz has said that both Bellow and Bloom "inhaled books and ideas the way the rest of us breathe air."", which is what I wish I had been able to do, sio perhaps I should make an effort, although the descriptions of his books have never really appealled that much, but perhaps I am plain wriong, though Bellow is, it is clear from the review, one of those grand old men of American letters who seem increasingly out of kilter with the world today. According to John Clute, Bellow is "generally thought of as the premier Mainstream novelist of his generation in the US", but that generation has very much passed now. I woulkd probably be much better off, for my own purposes, reading Fritz Leiber or more Gene Wolfe or Le Guin or Marlon James or even Robert E. Howard. So many books, so little time.

Of course, that cuts both ways. Yesterday I also came across (I had thought through some link from the Bellow article, but I can't reconstruct anything obvious to Walter Mosley's This Year You Write Your Novel. The reference I saw was to a quote from the book, although not anything inspirational, but toi osley's experience as a writer of colour. Well, I had to look up a book called that and it is exactly what the name suggests. It is another sign from fate. And priced, £3.99, to sell on Amazon. Of course, I make no assurances about which year is meant. 2019? The 366 days, 2 March 2019 to 1 March 2020? My 52nd year? My first year in Seattle? We shall see. But it must be, I know, I know, done. Before it is too late.

Timeline of the Gunpowder Age

According to the FoAK article Timeline of the Gunpowder Age, the earliest mention of what might be gunpoweder is in a Taoist text on alchemy dated to 142 CE, but it doesn't appear to be until the C10th that things start to getting going in China. Roger Bacon mentions gunpowder in 1267 (so the height of the High Middle Ages), which is the earliest mention of Europe in the timeline, so gunpowder was at least known as a curio amongst European academics at this time, although Bacon had probably never seen it in action, but just read it in a book. Gunpowder had reached the Middle East a few decades earlier. By 1326, we have the depiction of a cannon in Europe, but it is not until 5 April 1450 at the Battle of Formigny that triumph of gunpowder is fully realised with cannons demonstrating better range and rate of fire to the English crossbow. The classic gun emerges by the 1480s and remains largely unchanged for nearly three centuries.

So, what's interesting here is how long it takes gunpowder to become truly transformative. The Chinese certainly used gunpowder a lot, but it took them a long time to get going. The Europeans are bit quicker once they have gunpowder, but then probably learnt a lot from the Asians. One thing about gunpowder is that, as so often, it is just one technology in a set of linked technologies. Other examples include printing and sailing ships. Gunpowder is a low explosive, so it is only a lab curiosity until you stick in a tube that allows high poressures to be generated. Rockets are easier to make than cannons or handguns because you only a paper or bamboo or wooden tube to get a impressive effect whether just for entertainment or in comvbat. Cannons are serious bits of metallurgy, so you need to be able to make metal tubes that can withstand the stresses and temperatures of ignition of the powder. You also have to be able to make gunpowder on an industrial scale, which means you have to be able to obtain the raw ingredients on an industrial and store and transport the gunpowder safely and reliably. So, you can't just go from a laboratory demonstration of gunpowder straight to cannons blowing great chunks out of the castles of your enemies. You need a whole technoeconomic complex to go with it.

If you have a superweapon, you would use it. Well, wouldn't you? Obviously, there has to be a strategic and operational context in which you use the weapon. A medieval cannon isn't quite like a late C19th machine gun. Like a big seige engine, it looks aand sounds impressive and uis undoubtedly much more effective at blowing chunks out of castles, but it is only element in your collection of materiel. You are still going to have to rely on traditional weapons and tactics mostly of the time until handweapons and the classic gun become practicable. That goes change the nature of the battlefield as it is men with muskets and swords rather bows and swords than are fighting it out.

There is also the question of the local (in time and space) laws of war. Using gunpowder might have been seen as not playing fair. There are generally some cultural norms that govern wars within particular civilisations. Of course, when civilisations collide, things might take a very different course.

And is there anyway that the Romansd could have invented gunpowder? And if they had, could they have done anything with it they did not otherwise do?

 

The Lights in the Sky Are Stars

I remember thinking when I was a child that some of the lights in the sky were galaxies and thus contained millions or rather billions or even trillions of stars, but, of course, we can't actually see galaxies with the naked except for Magellanic Clouds and a hint of nebulosity in Andromeda under very dark skies if you have acute enough vision. Actually, I hadn't realised how big the Magellanic Clouds are until just now, another of those facts that one could have known decades ago. The LMC is perhaps a tenth of the mass of the Milky Way, so perhaps has 10 billion stars, although it is gas-rich and metal-poor compared to our galaxy. There is plentiful star formation going on and who knows what the true history of these galaxies might be. Yes, when we look at them we might be seeing civilisations.

I am sure I have made the point before that if we could see the true extent of galaxies they would be enormous. I am sure I have made the point also that if one were on an planet around an intergalactic star, one coulkd barely make out the nearby galaxies. But my original pooint was to ask what is a star? We can see objects that aren't stars. But is a neutron star or a white dwarf a star? I'd say it was. After all, a neutron star or white dwarf used to a star. Just because they are not undergoing fusion now, does mean they are not a star.

The term "fusor" (you see, there was a link to yesterday) has been proposed for objects in which fusion occurs at some point. That's fine for brown dwarfs (I do prefer the older term substar, but that's a lost battle) as they (mostly) are large enough to fuse deuterium for a period and some, I understand, even hydrogen. In fact, I suppose you could define a brown dwarf as a fusor that is not large enough to be a star (that is burn hydrogen on the Main Sequence), But I am not sure it is terribly plausible to say a neutron star is not a star, it is a fusor. 

The ternm planemo is hardly a elegant term. It stands for planetary mass object. So, anything smaller than a fusor, but large enough to have reached (more or less) gravitationally equilibrium is a planemo. A planet is then a planemo in orbit around a fusor. That would include, for instance, planets arounds neutron stars and white dwarfs. We need a term for planetary mass objects like the Moon and Ganymede that are not planets. I'm just not sure planemo is it. 

Farnsworth Fusor

I see there have been reports of a 12-year old boy in the US building a Farnsworth fusor. A slightly older boy did that in Lancashire a few years ago. I remember reading an article about them in Analog some years ago. You can build one with the kind of equipment to be found in an A-level physics laboratory. This is a good make and I wish I had a poersonal laboratory with the kind of equipment to be found in A-level laboratories at my disposal.

In the Farnsworth fusor, the deuterons tend to collide head on. This is energetically favourable for fusion, hence the usor is efficient. You will get plenty of fusion occuring. I wondered what was detected though. But deuterium does not fuse, surprisingly, directly to He-4, but to He-3 plus a neutron or H-3 plus a proton, so I guess you detect the decay of the tritium. But why doesn't H-2 fuse directly to He-4? Isn't He-4 especially stable (hence, after all, alpha particles)? What I hadn't realised before is that it is odd that tritium is not stable, despite having more neutrons than protons, but He-3 is and is the only stable nuclide with more protons than neutrons (excluding H-1, natch). Why is that, I wonder? It is again something I would like to know and w are not so far off technetium country (why does it have no stable isotopes - or should that be nuclides?), an issue I have wondered about here before. There is always more to known.

The Farnsworth fusor itself isn't going to give us electricity too cheap to meter, but perhaps it will be the Jonbar junction for the boy who built one and there are plenty of fusion start-ups out there (how many though are like Theranos?) It would be a neat thing to build in one's lab.

Superliminal and Time Travel in a Quantum Universe II

The fact that geeneral relativity allows closed timelike curves might be a clue that there is something wrong with general relativity. But a question is whether you can get CTCs in situations in which quantum effects will be insignificant? It turns out though that wormhole (Thornian) time machines always require negative energy simply in order to send spacetime in the required way. We don't have convenient lumps of exotic matter with negative and quantum mechanics suggests that if we did, they would be unstable. So this is what those Stanford Encyclopedia of Philosophy articles I was reading in the toilet at Imperial Consultants were saying. Though it looks like GR lets you create Thornian time machines, quantum mechanical effects will always (it seems) prevent them working in practice. Tiplerian time machines seem to be ruled out by the infinities involved with them, but, worse, work by Hawking and others seems to indicate that, for very general reasons, you are always going to need negative energy (or violate the null energy condition) to create any kind of time machine (or FTL mechanisms), which seems to rule them out.

That was very much the case in the 2000s and is the depressing conclusion that Everett and Roman reach in their book. There are few possible get-outs. Do the quantum equalities that rule out stable negative energy sources apply in every possible situation? (They are certainly quite general, but not necessarily absolutely universal.) Are there possible workarounds to Hawking's work (this is not his Chronology Protection Conjecture) that might mean that time machines are not ruled out generically? Are there Tiplerian time machines that atre not plagued by infinities? (What, for instance, of Gott's work on cosmic strings?) What about universes like Goedel universe? But his is a very toy one. And we don't seem to live in a non-toy version and even if we did, would it help? Would it be possible to make a universe like Goedel's? Of course, there is only a point in doing that if FTL/time travel is allowed. (Well, OK, if you can make and travel to pocket universes of your own design, there are lots of non-time travel/.FTL reasons for doing so.)

But let's go back to the fact that CTCs are allowed at all in GR. It's unreasonable to suppose that CTCs wouldn't be allowed in quantum gravity. That quantum effects seem to rule out certain kinds of time machines is suggestive. There is though no guarantee that this would be the case,. It's possible that QG might still contain CTCs and it is even possible that QG might override some of the objections to time machines in semiclassical gravity. It's probably too much to hope for, but we must still hope. Everett and Roman do make the point that, for instance, though it is hard for us to imagine what kinds of macroscopic sources of negative energy there might, there are examples of completely unexpected macroscopic quantum behaviour such as superfluidity, lasers and the Bose-Einstein condensate. So, it is quite possible that we could have all sorts of weird quantum gravity macroscopic effects that could be used to produce negative energy and that could then be used to create time machines or FTL drives if, for instance, the quantum inequalities did not apply. It's a long shot, but the best one we have at the moment.

At one point, E&R do vaguely gesture at the noytion of FTL systems that are not time machines. This would, of course, resolve problems like the grandfather and bilking paradoxes. The general position is that to preserve causality, if you have an FTL drive, you can always create situations in which, to some observers, effect precedes cause. That though is for some observers. I want to understand whether that situation is the same as a traditional time machine or necessarily leads to being able to create one. One can have a FTL and, Charlie-style, promise not to mess around with causality in your light cone - or else! But could you have within QG some way of achieving FTL, but automatically preserving chronology? As I understand it, SR says no, but then what are E&R getting at? 

I have been interested in the physics of time travel for years, enough ti read several books, but not enough - yet - to brush up on my SR enough to understand why FTL = time travel. After reading E&R, it does not look as though there is much point reading anything else on time travel or  FTL. But that would have the case 30 or 40 years too. But then we didn't understand a lot of the issue sio well. I do have Kip Thorne's The Science of "Interstellar" to read. That we can even talk about time machines or FTL drives scientfically is astonishing. Given that nature has tantalised us with its possibility, could it be so cruel to not let us find some way of achieving it eventually. And who knows what might seem obvious 20 or 200 years from now?

Superliminal and Time Travel in a Quantum Universe I

Further to my post on 31 December 2019, I have now finished Time Travel and Warp Drives: A Scientific Guide to Shortcuts through Time and Space by Allen Everett and Thomas Roman. I learnt a lot. Which is good.

In a Newtonian universe, there is no fundamental speed limit. (In modern terms, we could say that the speed of light was infinite, although, of course, we already knew in the C17th that it wasn't). But a ship accelerationing at 1g will still only have reached a speed of about 300,000 km/s after a year. So, it is still going to take you a long time to get between the stars unless you are able to tolerate very high accelerations. 

Special relativity though offers us a tantalising possibility: time dilation. If we can get close to the speed of light, time slow down. So, in a 1g torchship, we can get to arbitrary distances in a few years or decades of ship time. This is better than a Newtonian universe, at least to those on the ship. But, frustratingly, there's a catch. Relativistic effects don't really start hitting in until we are close to the speed of light. The amount of energy required to get to 0.9c, much less 0.99c or, particularly, for a continuing accelerating ship is pretty large. Where would you get this energy from? The vacuum energy would be my guess, but until we work out how do that, relying on time dilation to get us around the universe on semi-reasonable timescales seems dubious. But surely nature wouldn't be so cruel to hold up this possibility only ti snatch it away.

But, of course, we don't just have special relativity. What of general relatvity. Here the prospect are even more tantalising. The notion of the wormhole has been around for decades. The SFE claims it was coined by John A. Wheeler in 1957. There's obviously a link here to the concept of hyperspace. Wheeler was drawing on work by Weyl in 1928, who called his workmholes one-dimensional tubes. The term "hyperspace" was used by John W. Campbell in Islands of Space (1931), but had been used earlier by Kirk Meadowcroft in "The Invisible Bubble" (1928). Campbell's story uses hyperspace as we understand it today. By 1935, John Buchan could have a character who knew "...less about women than he knew about the physics of hyperspace." Where did Buchan get the term? Of course, popular descriptions of the fourth dimension had been around for decades, but Buchan's use of "physics" rather than "mathematics" is suggestive. Did he get the world from a young relative or an article or book on higher dimensions or general relativity. Ot had he read Campbell?

We have had some notions of using black holes as ways to get around the universe, but, famously though, it is not until Carl Sagan talks to Kip S. Thorne about Contact in the laye 1980s that the modern notion of FTL wormholes, grounded in general relativity, really gets going. What wonders what precursors there were to Thorne's work. 

An FTL drive is a time machine - or at least can be. I wish I could see that. Time Travel and Warp Drives is reasonably technical, but despite having a physics degree and, at one time understanding, to some degree, why time dilation had to occur, I just don't have the time, patience and desk these days to be able to work through the reasoning with pen and paper to understand fully why FTL = time travel. 

The immediate problem with wormholes is that, in order to bend spacetime in the appropriate ways, they require exotic matter. Now, there are lots of different types of exotic matter, but here is meant a specific form of very exotic matter: matter with negative energy. Now, we have plenty of lumps of positive energy just lying around, but not so much with negative energy. Perhaps we could make some somehow, perhaps with the Cassimir Effect. That was the thinking back in the late 1980s/early 1990s. 

Now, the Einstein field equation of general relativity also tantalisingly includes solutions that involve closed timelike curves, which are time machines. This has been known since the 1930s and Goedel in 1949 was explicitly looking for time travel in his rotating universe (hence why he kept asking Dyson if they had found it - evidence that the universe is rotating - yet). It was Tipler in the 1970s who popularised the notion that CTCs could be created by rotating very long pieces of matter at high speed. This seemed like a plausible route to a time machine.

A problem with Tipler's scheme is that it involves an infinitely long cyclinder of matter. Now if I have a cylinder centimetres or nanometres or femtometres wide and many AU or parsecs long, I might think it is near enough to be being infinitely long that I can ignore the end effects, but, sadly, no. It really does have to be infinitely long. 

But we don't live in an entirely Einsteinian uinverse. At some point, quantum mechanics will rear its (complicated|ugly) head. And I will talk more about the consequences of that in part II of this article tomorrow.

There's Always More

Consider this The Book Unbound, Theory of Clouds and Reading Time in the Twentieth Century, which is tutored by Caroline Edwards, who is an sf personage of whom I should have heard. There is so much interesting stuff out there. There can never be too much, but, this side of the Singularity, how does one even began to get started with any of it? How does one stop from feeling over whelmed by the sheer volume of things there are to be interested in.

These are all modules on the MA in Cultural and Critical Studies at Birkbeck. Of course, I don't live in London any more (what might there be in Seattle if I ever get there?) and might not qualify for the MA without doing an undergraduate arts/humanities degree first. As the Weasel has said "degrees come at degree time" and as Keith has indicated, why would you want to do a degree rather than just reading the stuff you want to? In a degree they make you study things you aren't interested in, you have to spend time around irksome people being irksome, you have to go to dull lectures and seminars, you have to write essays, and, worse, do exams, And it is all enormously time-consuming and expensive. And at the end of the day, most of the interesting things are not, for various reasons, as interesting as the idea of them. 

There's also an MA in the History of Ideas, which has the following introductory reading list

1. Berlin, I, Karl Marx: His Life and Environment (1978).

2. Berlin, I., (ed. Hardy, H.), Against the Current. Essays in the History of Ideas (2001).

3. Collingwood, R.G., An Autobiography (1939, with later reprints).

4. Collingwood, R.G., The Idea of History (1946, rev. ed. by J. Van der Dussen 1993).

5. Darnton, Robert, ‘Intellectual and Cultural History’, reprinted in The Kiss of Lamourette: Reflections in Cultural History (1990), pp. 191-218.

6. Dray, W.H., History as Re-Enactment: R.G. Collingwood’s Idea of History (1995).

7. Goldstein, J., Foucault and the Writing of History (1994).

8. Grafton, A., 'The History of Ideas: Precept and Practice, 1950-2000 and Beyond', in Journal of the History of Ideas, 67:1 (Jan 2006): 1-32. [Cut and paste this reference into your search engine – it is available for free as an open access article online].

9. Gramsci, A, Selections from the Prison Notebooks (Hoare, Q, and Nowell Smith, G, [eds. And trans.] 1996).

10. Kelley, D.R., The Descent of Ideas: The History of Intellectual History (2002).

11. King, P., Thinking Past a Problem. Essays on the History of Ideas (2000).

12. LaCapra, D., Rethinking Intellectual History: Texts, Contexts, Language (1983).

13. Skinner, Q (ed.), The Return of Grand Theory in the Human Sciences (1990).

Some of which would certainly be interesting. I am a big Darnton fan, for instance. The indicative option modules are

• Ancient Knowledge
• Britain and Germany: The History of a Relationship, 1815-1990
• China and the West: Encounters
• Civil Society: Ideas, Practices and Dilemmas in the Modern World, 17th Century to the Present
• Cultural History of War in Britain and America between the First World War and the Conflict in Vietnam
• Darwin, Darwinism and the Modern Ages
• Empires in Modern East Asia
• Gender in Britain
• Globalisation: A Short Historical Introduction
• Jews and Antisemitism in Modern Europe: Histories and Approaches
• Madness and its Meanings
• Magic, Science and Religion in the Renaissance
• Monastic Lives in Medieval England
• Politics and Islam
• Race and the Victorians
• Religion in Society and Politics: Britain and Ireland, 1801-2001
• Rethinking the Cold War
• Technology, Modernity and the Nation: Britain and Germany, 1880 to 1930
• The Making of Modern Societies: Britain and Europe, c.1500-c.1750
• The Rise of the Modern Consumer: Consumers and Politics in the Nineteenth and Twentieth Centuries
• The Two Irelands 1911-1998
• The Victorian Communication Revolution
• Venice and Istanbul, 1453-1797
• Vice and the Victorians: Moral and social reform in the late nineteenth and early twentieth century

So plenty to get your teeth into there.

Oddly, the MA in Modern and Contemporary Literature doesn't appear to offer the Reading Time in the Twentieth Century option, but the BA In English does have a module on science fiction.

Of course, what I really is an MA on historiographic metafiction, but I suppose you make your own (in that a MA is always going to be a lot about independent reading anyway and then there's the dissertation). Or, alternatively, technoecomic history. Or historiographic metafiction and technoeconomic history. It is not as though they don't overlap enormously. 

The Past Through Tomorrow

We know of different kinds of minds: human ones, great apes, cetaeans, bats, cats, dogs, New Calendonian crows, octopuses, social insect colonies, forests, human civilisation(s). We don't have to recapitulate the whole of terrestrial evolution in order to produce general intelligence as we have a pretty good idea what sorts of systems we want.

As I have said, five years ago, we might have predicted that a system that could play go at superhuman level might be 10 or 20 years off. But we got one unexpectedly quickly and in a way that was generalisable and that quickly led to Singularity-like behaviour with AlphaZero. John is quite right, Games are problematic for all sorts of reasons. Go no longer seems as ineffably human as it did half a decade ago. Now we demand - and right so - a Fields Medal or a double first in Mods and Greats or a symphony to rival those of Beethoven, Bruckner and Mahler.

AlphaStar "cheats" because it is so good at handling its units at a micro-level in a way no person could. Of course, again as I have said, the eschaton might well "cheat" and evolution doesn't have the stylistic concerns to go with victory conditions. We have no reason to assume that the AGIs would have minds very like ours at all even if we can talk to them. The Turing Test is a fascinating idea, but may well be more or less irrelevant to an out and proud AGI. Why would it want (or need) to imitate a person. It might well be beat at whatever it needs to beat us at by being able to handle its units at superhuman speed. 

I was listening to the Talking Politics podcast today (the episode with Adam Tooze) and saw that there was an episode with Martin Rees on existential risks. In the notes to the episode, Rees downplays the risk from machine learning. Human stupidity will remain or chief concern in "coming decades". But in 1919, no-one could have predicted how AGI might have worked, even though Henry Adams had already up with both sides of the Singularity (exponential growth and phase change). Neural nets come along in the early 1940s and then cybernetics and by the early 1950s Turing was writing his Mind paper and Ashby Design for a Brain. So, we don't know what the situation is likely to be in 2050 (perhaps 12 Moore's law periods or a 4,000x increase in computing power). From where we are now, it does not seem to me prima facie impossible that things could get interesting in the decade after 2025 or that all bets are off after 2035. (Yes, I woulkd feel more confident if more had happened with molecular nanotechnology over the last two decades.) But that's less than 20 years off. Of course, that long ago we didn't have Wikipedia or Facebook or Twitter or smartphones or AlexNet or deep reinforcement learning or IBM Watson (indeed) or Alpha* or the NLP revolution. 

But in the grand scheme of history, it doesn't really matter whether the Singularity happens in 20 years or 200 years or even 2000. It's all soon by whatever measure of historical time you care to adduce. Of course, we want it to be 20 years because I might still be alive. Which would be one of the most frustrating things. To miss out on the Singularity by a few yesrs or decades. It is is not as though the idea of the Singularity, of recursively self-improving systems, is going to go away. There are plenty of existential risks that might get me or the planet in the next few decades. But whatever the future is like, it won't be like the past. I think there is a novel in this. Wasn't Stephenson supposed to write it? (Sequel to Cryptonomicon. Perhaps he did in Anathem.)

Raw Compute

My first PC was a 25 MHz 386DX with 2 MB of memory and a 44 MB hard disk. That was October 1991. That was 27 years ago, so something like 15 Moore's law generations, which about a factor of 32,000. Now, you could run some pretty interesting programs on that machine such as the orginal versions of Civilization, Championship Manager, FractInt, SimCity, Wolfenstein 3D and Links. Certainly, Civilzation and Championship Manager had their longeurs, but the games were definitely very playable. Over the years, games have added better graphics, more features and faster load times, but also a lot of bloat of various different kinds.

I remember Steve Gallagher asking back in the day if there were some way of slowing down a PC game as it was too fast now he had a 386, but obviously something like StarCraft II would be hopeless on a 386, but it wasn't really a case that what would make Civilization better was it being faster, although that was always going to help a bit as would having better graphics. I actually preferred the 2D view to the isometric one that later versions use, but what Links be like now? Sadly, it is no longer produced. Are there any golf games for the PC? Obviously, golf games are something that can be constantly improved with higher resolution images of the courses and better game physics. Whgat would Links 2019 be like? 

Obviously, there are a lot of things that extra memory and  those extra compute cycles get used for. But would my 386 have been a supercomputer in 1975. And it wasn't as though people weren't doing useful things with computers then. But to train AlphaStar requires hundreds of years of time on a grunty TPU-based rig. NVIDIA offer the DGX Station, a personal AI supercomputer, which offers 500 TFLOPS. How much does it cost though? Price last March would seem to be about $49,900, so should be less by now, but, of course, the cost-effective of accessing such systems is via the cloud, although if you utilisation were high enough, it might be cheaper to have your own system. And obviously cooler. 500 TFLOPS would put the machine at fastest system in the world level in the late 2000s. But could something like this train AlphaCiv in a reasonable time? I can imagine a PhD student having full access, perhaps, to something like a DGX. That could also apply to researchers at AI-based startups. (If only someone would give us the cash!) It might well be that would need something gruntier than this for AlphaStar (although not to run the trained system). We might well get optimisations and libraries and pre-trained networks (transfer learning!), but ideally one would like to be able to able to train AlphaCiv from scratch. We've come along way since 1991, but will it be 2047 before we can do that a home?  

2019: Spring

It is not long ago that people wre claiming - I myself was claiming - that go was solved because go programs could play at at moderate club player level. After all, it would take several years to get to moderate club player level for most people. It was not long ago that people were claiming that OK, machines could now play a good game of go, but people could still kick their ass at games like Civilization, which involved hidden information and randonmess and long term strategies in a highly dynamic and heterogeneous environment, much less something like StarCraft II. which has all that plus is played in real-time.

But now we have AlphaStar. It is not quite true to say StarCraft II is solved, but the program can play and win at the grandmaster level against humans under competition conditions. There are ways of making things harder, such a relying just on what the camera can see on the screen or playing with different races on different boards (although humans tend to specialise in one race, I understand) It is not as though AlphaStar is not going to get better. As with those mid-2010s go program, AlphaStar can play a good game of StarCraft II and not, this case, just a good game, but a game at elite level.  

Is this a bigger deal than AlphaGo or AlphaZero? Arguably, it is because it is more obvious what you might do with an AlphaStar-style system. What about self-driving cars? Or financial trading? Or weather forecasting and climate prediction (an example DeepMind state in their blog article)? Obviously, for myself, I am interested professionally in anomaly detection in APIs. As Lex Fridman points out in his lecture I am currently watching on Deep Reinforcement, it is not quite right to say that DRL is not supervised learning. It is supervised in the sense that someone has assigned a reward function. In traditional supervised learning, that's how many predictions the system got right. With a DRL that could still be the case (did it predict or postdict the weather correctly?). but usually the reward is something related to winning the game. Of course, this doesn't have to just be one point for a win. AlphaStar has a league table and lots of different variants specialised for certain strategies. A backgammon or poker or trading sysyem would care not about winning as such, but how much money it makes. A problem for my AlphaAPImetrics is what exactly is an anomaly? I need to think about this. Perhaps there is some way of defining a reward in the system. 

So, now that we have AlphaStar, does that mean that all of the Atari games are solved? What of backgammon, bridge, poker? Is AlphaCiv now just a weekend hack for undergraduates? Exactly how much raw comouting power would you need for an AlphaCiv. AlphaStar trained for something like 200 years. Clearly, there is still a lot of improvements yet possible to reach human-level in terms of speed and efficiency of learning. Of course, that might not matter. What would happen if you turned AlphaStar up to max? It makes a comparable number of moves per second to a human player, but that's partly because it is deliberately throttled to avoid the agent winning simply because it can coordinate large numbers of units at superhuman speeds. Of course, there is no guarantee than a(n un)Friendly AGI would do that. 

Deep Learning is a sexy name, but it is really just using bigger neural nets and more data. You couldn't do that in the 1980s or 1950s. OK, there are some tweaks in terms of the architecture to get good performance out of the system, but there can be feeling that just getting better at face recognition is an ability with diminishing returns. There are only so many situations in which you have large collections of labelled training data. DRL on the other hand does seem more "magic", if there aree domains to which it can be applied. People might get bored with just deep learning, but combining AlphaStar and BERT suggests to me that there might be more some mileage in the AI hype cycle yet. As I said before, A(G)I isn't going to stop being a good idea. It's a question of finding the right problem to solve. We can already see some of those questions. Bu what we need is a system capable of getting a first in maths at Oxford. That's a DRL problem. It will be interesting to see how long it is before that one is achieved. And as long as the research dollars go into that kind of problem, I don't think we will see an AI autumn.

Seahaven Towers Revisted

Revisiting my "Seahaven Towers" entry from April 2007. I am struck by a jolt of jamais écrit. Did I write that? Not just in the sense of this "This is terrible" or "My God! This is actually good. Did I write this?" but also the sense of "Wow! It's hard to believe my mind was still sonimble then it could produce something like this!" I'm 50, nearly 51, and, perhaps, a big perhaps, I might get my visa and get to have a new start in the only the second city in the US to be named as a UNESCO City of Literature. My Forties weren't the decade of writing they were supposed to be and I am nearly a tenth of my name through my fifties. The years and decades rush by and yet this dispatch from 12 years ago seems to belongs to another world. I remember flashes of it, but I was trying the other way to remember what it must have felt like being me in that epoch and I just can't do it. Obviously, in a lot of ways, I did feel much as I did know, but there are a lot of details of everyday life that are lost. In April 2007, I was still in my first year at Wallingford Software. I still had nearly another years of commuting from Brentford to Crowmarsh Gifford, a 95-mile round trip. And for some of the later period, I was travelling from or to Wanstead in the morning or evening. 

I remember doing to see Love Actually, so probably 2003(!), at the Barbican and A saying she liked it and me opining that that was because we were living in a romantic comedy. Or, in 2000, being on a Routemaster with V, right at the on the lower deck and you are looking straight ot through the windows and it feels like you are flying. And, of course, that's just a capricious moment that can never be recaptured and is gone in a minute or two and nothing as good as that can ever happen again. 

Part of the problem is that things are only ever really good in hindsight. This is the problem with having fun. At 50, what can you do that's fun? As the Weasel said, wherever you are, that's where you are. And yet I did enjoy Seattle when I was there and do want to go back. It's good being able to set off forthe Elliott Bay Book Company at 18:00 on a Sunday because it is still open on Sunday evenings. You can't do that in London, much less Preston. So, I'd like to go to the Olympic National Park and the Mount Rainier National Park. I'd like to visit Hurricane Ridge and Paradise Inn and Sunrise Lodge. The idea of something is better than the thing. The idea of going on a holiday is better than the holiday itself (hence all the adverts). The idea of having been on a holiday is better than the holiday itself. The existential contingencies of actually being somewhere mean that it is never as much fun as the idea of being somewhere. Of course, a lot of that is that the memory lasts, but the moment doesn't. I can remember a lot of moments, it's all the other sdtuff I can't. and some of it I definitely would like to. 

Which is one of the things. I have been doing this blog since 2004. Blogs are hardly what they were then, but there is no reason to stop. Where will we all be in fifteen years? Will we even still be alive. 15 years can't be a long time, but being at Siemens might as well be another world. So in that sense there is a lot of time between now and 2034. Enough time to write a book. Especially with the help of Hugo House.

In other news, PySol is, as the name suggests, a Python solitaire collection. That surely has an API. Well, you would hope. So might be a possible opening to attack with AlphaZero.

AlphaZero and Seahaven Towers

I understand from the second MIT A(G)I lecture (there appear to only be two online at the mopment, although presumably ten have already happened) that one of the breakthroughs in 2018 was an emphasise on training models quickly. A problem with a lot of ML applications is that you need a fairly or even very grunty system to train your system. Which is fine if you are Facebook or Google or probably a PhD student at MIT, but so fine if you are undergraduate at MIT much less one at UCLAN or me. So there's been a competition about to train models in the minimum time or with the minimum cost.

Of course, you can use a pretrained model and tweak the parameters a bit, which is fine for some applications, but not so good if you want to do something more novel. This doesn't really help me, of course, with a six-year old laptop, but at some point I hope (have, indeed, been promised) something a bit more up-to-date.   I believe our friends at fast.ai have been at the van of improving the efficiency of training models. I don't know how much this would help though with with non-supervised learning tasks such Deep Reinforcement Learning.

What I'd like to do is to implement AlphaZero for FreeCell as a a Proof of Contect. I have seem an example of being this done for Connect Four. A quick look online seems to indicate a number of people have tried this, so that's something to keep me busy looking into that. The question is how powerful a machine does one need. FreeCell would be an interesting game to have a go at. Doesn't look like anyone has done that. Indeed, one of the hits is to an Atomic Razor post from last week.

Unlike Connect Four or go, there is no opponent, so it should be easier, but, of course, there are a lot of different deals to deal with. I feel that it should be fairly quick toi train an AlphaZero-like system to become good at FreeCell. And issue is preventing the system just memorising all the games, but there are a lot more deal than the Microsoft 32,000. I play a game of FreeCell a day. Generally one is enough. I have been playing it on and off since those days of Seahaven Towers on Silicon Graphics Indigo workstations inside the reactor hall at the ILL in 1995. I wanted to write a What a Carve Up-style novel called Seahaven Towers.

Actually, Seahaven Towers is similar to, but not identical to FreeCell. Would a system that was good at FreeCell only reqire tweaks to be good at Seahaven Towers. Would transfer learning be possile here? I feel I have got better at FreeCell and developed a certain style. Would the program have its own style? Would the style transfer across games? Does AlphaZero play chess, go and shogi in the same sort of way.

It might be possible to create an AlphaZero solitaire player with limited resources. But what of something like backgammon or bridge? These are stochastic and, in the case of bridge, a game of hidden information. As far as I know, AlphaZero doesn't do well there, much less with Civilization or StarCraft II. Those, sadly, are not going to be problems that I personally am going to crack. But if only there were an API for AisleRiot Solitairwe. Perhaps I could somewhere with that.

 

AlphaZero and Seahaven Towers

I understand from the second MIT A(G)I lecture (there appear to only be two online at the mopment, although presumably ten have already happened) that one of the breakthroughs in 2018 was an emphasise on training models quickly. A problem with a lot of ML applications is that you need a fairly or even very grunty system to train your system. Which is fine if you are Facebook or Google or probably a PhD student at MIT, but so fine if you are undergraduate at MIT much less one at UCLAN or me. So there's been a competition about to train models in the minimum time or with the minimum cost.

Of course, you can use a pretrained model and tweak the parameters a bit, which is fine for some applications, but not so good if you want to do something more novel. This doesn't really help me, of course, with a six-year old laptop, but at some point I hope (have, indeed, been promised) something a bit more up-to-date.   I believe our friends at fast.ai have been at the van of improving the efficiency of training models. I don't know how much this would help though with with non-supervised learning tasks such Deep Reinforcement Learning.

What I'd like to do is to implement AlphaZero for FreeCell as a a Proof of Contect. I have seem an example of being this done for Connect Four. A quick look online seems to indicate a number of people have tried this, so that's something to keep me busy looking into that. The question is how powerful a machine does one need. FreeCell would be an interesting game to have a go at. Doesn't look like anyone has done that. Indeed, one of the hits is to an Atomic Razor post from last week.

Unlike Connect Four or go, there is no opponent, so it should be easier, but, of course, there are a lot of different deals to deal with. I feel that it should be fairly quick toi train an AlphaZero-like system to become good at FreeCell. And issue is preventing the system just memorising all the games, but there are a lot more deal than the Microsoft 32,000. I play a game of FreeCell a day. Generally one is enough. I have been playing it on and off since those days of Seahaven Towers on Silicon Graphics Indigo workstations inside the reactor hall at the ILL in 1995. I wanted to write a What a Carve Up-style novel called Seahaven Towers.

Actually, Seahaven Towers is similar to, but not identical to FreeCell. Would a system that was good at FreeCell only reqire tweaks to be good at Seahaven Towers. Would transfer learning be possile here? I feel I have got better at FreeCell and developed a certain style. Would the program have its own style? Would the style transfer across games? Does AlphaZero play chess, go and shogi in the same sort of way.

It might be possible to create an AlphaZero solitaire player with limited resources. But what of something like backgammon or bridge? These are stochastic and, in the case of bridge, a game of hidden information. As far as I know, AlphaZero doesn't do well there, much less with Civilization or StarCraft II. Those, sadly, are not going to be problems that I personally am going to crack. But if only there were an API for AisleRiot Solitairwe. Perhaps I could somewhere with that.

 

The Aliens Around Us

As I have said before, for all we actually know, the Solar System might be teeming with life. The clouds of Venus, the aquifers of Mars, the lakes of Titan, the clouds of the gas gianys, the subsurface oceans of all those satellites and asteroids, the geysers of Enceladus and Io. We simply haven't explored these environments to any significant degree. They might be full of microbes or even, in some, macrobes. 

Of course, what goes for the Solar System, goes literally billions of times over for the Milky Way galaxy, which is crammed with every imaginable time of environment, including ones much closer, in various ways to the Earth itself. So, it is is indeed, the case that there could be a lot of life out there. Of course, we know, from our own evolutionary history, that getting from single-celled organisms to multi-cellular ones to general intelligence to tool use to fire to language to civilisation to civilisation capable and sending and receiving signals across interstellar distances to civilisation capable of colonising the galaxy is non-trivial.

We iusually say there are no obvious signs of large-scale engineering works in the universe. We look out at the universe and we are not in general struck by phenomena that are best explained as the products of alien civilisations. Pulsars are rotating neutron stars, for instance, not interstellar beacons. But what if we are just looking in the wrong places? What about fast radio bursts? We don't really have a clue what they might be. We do know now that they can repeat, which would seem to rule out a lot of explanations in terms of events like explosions caused by merging objects. I suspect, as with Tabby's Star, that a convincing natural explanation will eventually be adduced, but nothing looks terribly promising at the moment. Fast radio bursts could be artificial in origin, although how we woulkd prove that for certain is another question. They might, for instance, be microwave beams for driving interstellar sails. Interstellar sails like ʻOumuamua? Again, ʻOumuamua is probably a natural object, but we simply don't know. It is very very anomalous. It is like nothing we have seen before or predicted. For instance, ʻOumuamua is virtually at rest with respect to the local inertial frame (the average motion of nearby stars). This is not to be expected and indicates that it could be a space buoy. In effect, it was less of a case of  ʻOumuamua rushing past the Solar System as the Solar System rushing passed  ʻOumuamua. And there might be other interstellar objects flooding through the Solar System. It would be great if we were able to mount a mission to ʻOumuamua. Whatever we found out would be amazing and there is certainly, at the moment, a non-zero chance that it could be alien. The aliens could be all around us.