Atomic Razor

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. 

The NLP Revolution

The Lex Fridman's MIT deep learning course has started again. I knew about this last year, but didn't get to see any of the videos, but am determined, as part of my New Year resolutions for 2019, to make an effort to watch them. (I am still hoping for another tranche of Andrew Ng courses.) 

Deep learning has been around for a while now and, of course, neural nets have been around for decades. Deep Learning just involves bigger neural nets and more data. Obviously, there is more to it than that, Deep Reinforcement Learning, for instance, is quite a different beast in a lot of ways, but it would be easy to feel that the Peak of Inflated Expectations had been reached. It would be easy to feel that if it were for what actually happened in 2018, which was the NLP revolution.

I have to admit that I was only vaguely aware of this. I had seen a few mention of it and hints about things around it, but I have not been paying the assiduous attention to the AI newsletters to which I subscribe. Again, that's something I can't let go in 2019. Various improved machine translation systems have been around for a while and they do seem to be getting better as far as I can see from translating Wikipedia pages. Certainly, the translations, if not quite perfect, are much smoother and more idiomatic than was the case only a year or two.

But the hot new thing is Google BERT. I hadn't even heard of it until last night, but apparently it created a splash back in November. It is open-sourced and already other groups have incorporated it and related ideas into their own systems. This is one reason that everyone has been excited about digital assistants. These have been around for a while now, but are coming on, I understand, in leaps and bounds. See, for instance, the leaderboard for "the very challenging GLUE benchmark, a set of 9 diverse Natural Language Understanding (NLU) tasks" on which BERT is now behind a Microsoft project. BERT reached human-level on the Allen Institute for Artificial Inteligence SWAG dataset, although it wasn't specifically optimised for the dataset. A problem with a lot of A(G)I problems is producing a system that works well for more than one problem. With Kaggle competitions and their like, there is a danger of overoptimising to the particular problem at hand. It's not that the system just memorises the training set, but that it is tuned to recognise just fish, say, and then has to be retrained expensively to recognise anything else. But with AlphaZero and BERT, we are definitely seeing transfer learning at some significant level. 

I do need to start exercising Google Assistant a lot harder. How good is it now? Now, I'd like to have a play around with these kinds of systems. A problem is the sheer amount of compute needed. I have a six-year old laptop. Yes, I need a new one, but how far can one get really get with our a fairlygrunty GPU or even a TPU? It would be nice to apply an AlphaZero-like system to something like FreeCell. But do I have the compute leaving aside whether there is an API for AisleRiot Solitaire?

A interesting facet of the NLP revolution is that there had been concern that had performance on various Deep Learning tasks had been plateauing, such as in computer vision or self-driving cars, but then we see sudden breakthroughs as also with AlphaGo/AlphaZero. But it is also clear that the possibility of rapid or even very rapid progress over short timescales is also possible. It is 2019 and I still think there is no reason to think that beyond 2025, we might see the emergence of true AGI. That 03:14:06 UTC, 19 January 2038 Singularity seems still very much on the cards.

 

Information Overlord

Some days there is just too much going on in my head

• Cybersyn
• Eden Medina
• cybereconomics
• Corbynomics
• "cyborg" socialism
• John Cruddas
• Beer's and Pask's pond
• Brexit
• Inventing the Future
• Fully Automated Luxury Communism
• B*st*n*
• Sir Geoffrey Vickers VC
• Systems Theory
• the Lorentz transformation
• the absolute centrality of MIT Press books
• my library of Verso books
• all the other books I have
• applying for a US visa
• Python
• language in Alsace
• Dry January
• dealcoholised wine
• that irritating new Economist advert
• Game of Thrones
• True Detective (season 3)
• places to eat in Preston that are open on a Monday
• running
• virue as its own reward
• the new Space Race
• artificial (general) intelligence
• the best place to live in 2100 (I will be 132)
• the size of the world
• too many things
• not enough things.

Earthrise Over Compton Crater

With all the excitement about the recent landing of the Chinese rover on the far side of the Moon, I hadn't realised that the US Lunar Reconnaissance Orbiter probe has been orbiting the Moon for nearly a decade now and has created high resolution maps of almost the entire surface excluding only the polar regions in deep shadow. It has done particularly resolution surveys of the Apollo landing sites.

One image it returned, not a map, but the sort of thing you might see if you looked out of the window of a spacecraft orbiting the Moon.

(Source: By NASA / Goddard Space Flight Center / Arizona State University - http://lroc.sese.asu.edu/posts/895, Public Domain, https://commons.wikimedia.org/w/index.php?curid=45765646.)

As the Wikipedia page discussing the image makes clear, it might not look exactly like that if you were there. (Just as Mercury is not blue despite what the usual photograph of it seems to imply.) Nevertheless we can take it from Apollo astronauts that Earthrise over the Moon is pretty spectacular. No-one has seen that for more than 46 years. 

SpaceX had planned to do a Zond-style free return mission in 2018, but that got shifted to a BFR trip not before 2023, which probably makes more sense. I do seem to recall some suggestion that Blue Origin might have considered a similar flight. Yes, it would be a bit of a gimmick. more of a gimmick than Apollo 8 in 1968, which at least crew-rated the Saturn V and tested a lot of the systems needed for the later lunar flights. But it would be a hell of a trip. I wonder if there could be the possibility of a Blue Origin flip with the New Glenn. Could Bezos resist?

I might be nearly 51, but as my mum said to me today, my life is half-done. I have another fifty years to go. (I hope). One day, I would like to see Earthrise over Compton crater withmy own eyes. It could even happen.

Shortcuts through Time and Space

I was in the main library in Brighton the other day (S needed to print something out) and I spotted the Stephen Baxter section. From what I can tell the sf books and other genres are mixed together with the mainstream fiction, which may or may not be a good idea. Perhaps there is an sf section somewhere. A problem with libraries these days is that they are not as book-orientated as they should be. Anyhoo, it is difficult to keep up with Steve and I picked up a copy of, I think, Xeelee: Vengence. It was, I don't think the most recent "Xeelee" book from 2018, but the 2017 one. It had an acknowledgements section at the back and we all know the paratexts are the most interesting thing in most books and certainly the ones that people are most likely to heed casually flicking through a book plucked from the shelf.

In the acknowledgements, the first book Baxter mentions is Time Travel and Warp Drives: A Scientific Guide to Shortcuts through Time and Space by Allen Everett and Thomas Roman from 2012. As you know, I am extremely interested in semi-popular books about time travel (and FTL, which is the same thing in a relativistic universe) and I had somehow managed not to come across this one. before so many thanks to Steve for putting me on to this. Luckily, the book was available on Kindle so I was able to bag it straight away. It also solved the problem of what to read next, which had been bugging me in the hotel in Brentford the previous night. I hgave a lot of books on my Kindle, a lot of good books as pretty much all of them are books I have chosen personally, but which do I actually want to read now.

The book is slipping down, although I doubt it will be as good as Paul J. Nahin's works on time travel, not even referenced in the current volume. Of course, what I really need to do is to coinvince myself that FTL requires time travel in a relativistic universe as a physicist who did study special relativity to a certain degree as an undergraduate decades ago. I hope Everett and Roman will be a help here. They say they are going to be and the book is aimed at a undergraduate audience. But, of course, I really need those afternoons with pen and paper in my study.

The thing is though that the book is a University of Chicago title. As you know, they give away a free title each month. It would be just my luck if this book is given away this month or before I have finished reading it. There is a probably a time travel parable in there somewhere. 

It is very easy to come up with whole classes of book I ought to read in 2019. But I have a couple of books I got recently I need to finish off (the new Posy Simmionds and Fantasies of the Library) and now the Everett and Roman, so that will keep me going for the next few days. But it doesn't help me get nearer to becoming the new Stephen Baxter does it?

The Size of the World

I just lost my post about the size of the Traveller universe. Typepad: it is nearly 2019. Some kind of autosave feature is not merely desireablm, but close to essential.

OK, my various were that our world is big, but the Traveller one is much bigger in lots of ways (size, time, population, technology, aliens). And the Traveller universe only occupies a small fraction of the Milly Way, although it is implied that there is something unusual about our part. Hints have been dropped about what might out there, but only hints. 

The Imperium is highly decentralised, but has a common currency. How is it used as a instrument of Imperial policy? Is there an Imperial Reserve Bank on Capitol? How are interest rates set, especially given the time lags between core and periphery? Are credit coins physical realisations of some kind of quantum entanglement system with the complementary particles stored in the secure facilities of the central bank? Does this make the coins impossible or very difficult to forge? And how do credits fit with cryptocurrency systems? This is not something that could have been imagined in 1977, but in 2018, we demand answers. As I said, people are not the same across space and time. Someone in C21st CE does not - cannot - possess the same worldview as someone in C21st BCE urban China or rural Britain. We don't even possess the same worldview as 1978.  

Perhaps Orion's Arm is something like what I am after, but who has sketched out a space opera environment in the greatest detail or at least given systematic thought to some of the interesting features implicit and explicit in these milieu? Presumably, there are explored to somne degree in the New Space Opera, so writers like Vinge, Banks, Ken MacLeod, Reynolds, Baxter and probably lots of others have played in this sandbox. But is there anyone who has done it canonically well for the late 2010s?

 

Hollyhead Fold

A map of the late 1880s shows a Hollyhead Fold, which I presume is a farm in the vicinity of Hollinshead Fold Bridge on the Lancaster Canal. I have also learnt that Hollinshead is a name of a prominent Lancashire family. The question is whether the map is correct. Perhaps the surveyor was told the name as Hollinshead Fold, but misheard it. Or perhaps the farm was called Hollinshead Fold in the C18th when the canal was built, but was Hollyhead Fold a century later. I suppose it is possible that the bridge name is a corruption or over-correction of the farm name, but that seems less likely. 

Interestingly, unlike some earlier maps, this mapdoes not show a Cadley Moor or a windmill. I presume there was a watermill on Savick Brook on Mill Lane at one time. In the late 1880s, it would still all be fields around here for another four decades. Which fields belonged to which farms would vary over time. Cadley Moor might have been coomon land. Like Preston Moor, which had been Moor Park from the 1830s and the first municipal park in the Nritain, although not a planned one as Birkenhead Park was, although it became one, the moors would never have terribly moor-like as they are neither high nor hilly. The FoAK indicates that the land on which Moor Park was established was a farm., although it was originally common land. Indeed, there are plenty of edgelands around Haslam Park and Sharoe Brook still, only a few hundred metres from here, which give an indication of what the wider area must have been like in earlier centuries.

From Hollinshead Fold Bridge to Ingolashes Bridge

I've walked before about by desire to travel by canal from Woking to Preston via Brentford. The online maps I could find weren't very helpful in indicating the route. In Skipton the other week I stumbled on a copy of the Collins Nicholson Waterways Guide 5 North West & the Pennines. This is helpful in indicating how exactly having got onto the main branxh of the Leeds & Liverpool Canal via the Ribble Link and the Rufford Branch of the L&L, one can get to into, I think, the Bridgewater Canal and from there, via Manchester into the either the Trent & Mersey or Macclesfield canals towards the Midlands. (Or, alternatively, if one were coming from the south, the reverse.) 

The Nicholson guides originated in the 1970s. Mine is the 2006 edition. I bekieve the most recent is 2015. Many of the pubs listed in my copy will have disappeared now or changed beyond recognition. The Lime Kiln, the Dr Syntax and the Fylde Tavern are long gone. The Hand & Dagger, on the canal, outside Preston is very much still there. We aree going tonight.

One interesting thing though that guide provides is the names of the bridges on the canals. On the Lancaster Canal, 11A is Blackpool Road Bridge and 12 is Woodplumpton Road Bridge. 13 is Savick Aqueduct. 14 though is Hollinshead Fold Bridge and 15 Ingolashes Bridge. I have crossed both of those hundreds, perhaps even thousands of times and never had any idea what their names might be or even if they had names. Hollinshead Fold Bridge is the one you cross if you are going the quick way to Haslam Park. I think there might be a Hollins Road or some such nearby. Yes, a Hollins Grove. The bridge even got a namecheck on the BBC News website in relation to the amount of duckweed on the canal this year.

The names might well appear on OS maps (the guide includes extracts of OS), although my ones of Preston are in storage. So I could have seen them on threre. Perhaps there are people who know and use the names rather than "that bridge that you go over to go to the park" or "that bridge that goes to that field that goes to the park." Perhaps they are used by the Canals and Rivers Trust and in other official documentation.

The area has obviously changed quite a lot since the canal was built in the C18th. Hollinshead Fold sounds like a feature on a farm. This part of Preston is a classic edgeland. It would all have been farmland round here into the early C20th. The field that you enter when you cross Ingolashes Bridge, which I do when I go on my run, is part of the valley of Sharoe Brook. The path goes into the park and the field is now in fact part of the extended park. I don't know when it became part of the park rather part of a farm. Perhaps it has long been an edgeland. That's the kind of detail that it would be fasinating to be able to back in time and find out. I don't recall things around here changing much over the last 50 years, but surely they must have in all sorts of ways. There are certainly private fields around the park and even the foundations of a ruined house. I remember the house, quite substantial, when it was still standing, so there must have been farm fields around here well into my time. The Ingol council estate wasn't built until the 1960s, so that must have farms until then aroud the east side of Ingolashes Bridge. There are some 1930s houses there too. I live in a 1930s house too and it is hard to imagine what it must have been like when it was all farmland/edgeland around here a century ago. But the canal and the railway would have been here then. Go back another century and the canal would have been and the bridges. But it must have been quite different to today. It would be a sight to see.       

(A quick Google suggests that Bostocks Farm was the name of the farm the fields of which were incorporated into the nature reserve in the 2000s. I don't know if the runined house was the farmhouse and I have no recollection of the land being heaily worked, but perhaps there were sheep or even cows. It would be good to have better recall of things like that. This suggests that the ruined house was Ingol Lodge. The house was listed and burnt down in the 1990s and was then demolished, although the remains of the foundations were, last time I was there, very extensive.) 

 

Occupied Yorkshire

I was in Clayton-le-Moors last weekend, which was why I was not blogging. We went to Skipton on the Saturday. This was probably the first time I had been on the A59 beyond Gisburn since 2004. Gisburn is in Ribble Valley and one thing I hadn't realised is just how of occupied Yorkshire there is after Gisburn. You actually go into Pendle before you get to Yorkshire. I couldn't spot if the administrative border is marked. There are some suggestions I have come across that it might not be. Lancashire County Council might fear vandalism from Yorkshire irredentists.

There's a post the Ranter did a while back in which he describes the borders of Yorkshire. The Ribble is unusual in that it is the only large river that begins on the Yorkshire side of the Pennines, but flows west. The Ribble Valley thus creates a natural conduit through which Yorkshire could pour into Lancashire. And pour it did. On the north bank, Yorkshire goes up to the Hodder and on the south bank to only just east of Clitheroe. the villages close to Clitheroe, like Waddington or Great Mitton or Dunsop Bridge, were nowhere near anywhere in Yorkshire. One could argue these were really bits of Lancashire that were accidentally in Yorkshire. That's less obviously true for somewhere like Gisburn. It is slightly nearer Clitheroe than Skipton, but the inhabitants might have preferred to look east. But the boundaries of the Bowland Rural District were unalienable.

That's part of the issue. There might be Yorkshire nationalist who dream of a two-sea polity. At its westernmost point, the Bowland salient can't be much more more than ten miles from the Irish Sea at Glasson Dock. But I don't want any part of Lancashire that is not in Lancashire. There is, for me, no desire for a Greater Lancashire. "One day, this will be Lancashire!" No, Occupied Yorkshire should be returned to Yorkshire and we should all those parts of Lancashire annexed by Cumbria, Yorkshire, Cheshire, Merseyside and Greater Manchester.

The thing is though, which borders? That's easy, you might, say. The 1888 borders. But borders change all the time. The course of a river or strteam might change. A farmer might buy a field. Where's the border now that the fence or wall is not longer there? It's all just a field. A town might grow across a border. It may make little sense to have one part administratively in another district to the main part. A friend of mine lives in Shipston-on-Stour. Shipston was/is in an exclave of Worcestershire surrounded by Warwickshire. That anomaly survived 1888 and was rationalised away in the early 1930s. I address his Christmas card to Worcestershire, but should I really? Warrington is administratively in Cheshire because if it weren't, it would be an exclave of administrative Lancashire. Southport could be in Lancashire. Is it really a part of Merseyside in a demographic sense? WAs it grandfathered into Merseyside on the basis of unalienable 1888 boundaries ultimately derived from those of the ancient hundred of West Derby? 

In Engel's England, Matthew Engel visits Barnoldswick. (I wonder if I have that book. I might have got from the Chiswick remainder shop.) The town was largest place to be transferred from Yorkshire to Lancashire. If there is anywhere that could most properly be called Occupied Yorkshire, it is here. Though it can't get television from Winter Hill, it could be argued to be more in Lancashire than Yorkshire. Again, it's not really near anywhere in Yorkshire, but it is near Colne and Nelson. Engels asks people whether they think Barnoldswick is in and should be in Lancashire or Yorkshire. Naturally enough, they are suspicious of him and tend to give the politically correct answer. To most people, it doesn't really matter, of course. The maternity facility for Barnoldswick is these days in Burnley, but there's no birth qualification to play for Yorkshire any more anyway. Some people might prefer the sometimes Labour LCC in Preston to the inevitably Tory NYCC in distant Northallerton (what of Pendle itself though) and a teenage girl might opine, in disgusted terms that only elderly people think Barnoldswick is in Yorkshire.       

I share her distate for Yorkshire naturally. It shouldn't be the case that Lancashire is like Yorkshire only better, but rather that Yorkshire is like Lancashire only worse. Of course, that is just a bit of goodnatured banter. I come from Lancashire and my intonation goes down at the end of sentences. No ad voiceovers for me. Yorkshire is a foreign country and I hate most people from Lancashire far more than I hate any Yorkshire person.

Of course, the kinds of people who dream of restoring 1888 boundaries are going to largely be the kinds of people who voted for Brexit. So what am I do here really? Shouldn't I be siding with the liberal metropllitans elites drawing lines on maps to maximise administrative convenience? But as Kevin Rattan said many years go, if you live somewhere like Waddington and have been happy in Yorkshire for centuries and then that is taken away from you, it is upsetting. Lancashire as we know today did not really emerge until the C12th. When did the boundaries become something close to the 1888 ones, Barnoldswick was claimed at times by Blackburnshire and when did the county become more important than the hundred. Bowland, for instance, crossed the border and did the few who lived there care which side they were on or did they think of themselves as living in Bowland first? And 800 years from now, what lines will there on the ground in this disputed territory?  

           

Occupied Lancashire

Back in the old days, I used to only read short books, ones less than 200 pages long, in November and December in order to get my total for the year up. There's only any real point doing that if you are keeping count and a list (indeed), although I did get to read some curios I otherwise might not have done.

Joe Kennedy's Games Without Frontiers I mentioned a while ago as the book that I saw in the Elliott Bay Book Company and clued me into him. It would probably be less than 200 pages. Having read Authentocrats, I thought I might as well make 2018 the Year of Kennedy Completeness. 

Early on (I am only a third of the way in so far), Kennedy describes a trip to Barrow to see a game against Darlington. We get this sentence

Finally working my way upstairs to the bar, wide windows looked out across the muddy, balding pitch and let in the watery sky of Westmorland.

I suppose it is just possible, depending on the relative position of the windows that some of the photons entering Joe's eyes originated above Westmorland. It is equally possible that none at all did. We can though, from a glance at the map, be certain that the vast majority did not. Barrow-in Furness, of course, is historically, traditionally, in Lancashire and that watery sky was the watery sky of Lancashire. I know it well.

That "the watery sky of Westmorland" is more poetic than "the watery sky of Cumbria" does not stop it being wrong. Kennedy in trying to be poetic and possibly trying to demonstrate some slightly arcane geographical knowledge committed a solecism. I don't know how many Lancashire irredentists there are in Barrow-in-Furness, much less amongst supporters of Barrow A.F.C.. I do not know how enthusiastically Lancashire Day was celebrated there last week. But I'm a Lancashire irredentist. Administratively, it seems unlikely that the Furness and Cartmell peninsulas - North Lonsdale, Lancashire North of the Sands - will be returned to Lancashire. But we can at least hope that more can be done to remind people that these places are in perpetuity proper parts of the True Lancashire, something even poetically minded writers of cultural analyses of fottball should bear in mind..      

   

Infinite City

If you entered Preston from the west along Blackpool Road (A5085) and continued on to the junction with the A59 (which is just a small roundabout surprisingly) and then continued down New Hall Lane for a little bit until the end of the obviously built-up area, the distance you would have travelled is about 5 and a quarter miles. It would take you about 15-20 minutes to drive the distance (there are a lot of traffic light) or about 80 minutes if you walked - or longer if you stopped off for some liquid refreshment on the way. There are at least three pubs directly on the route plus one more that you can see just down a side street. From my house, you would have to add another mile to get on to the route and back, so an entire psychogeographical dérive would have a distance of about 11 and a half miles door to door. I could walk that in three hours. (I suppose technically speaking it might be a dérive if you just walked along it having planned to do that, but then how is that any different from deciding where to go based on any other arbitrary rule?)

Preston is wider than it is tall, so this rute gives you more of Preston than you would get going north-south, for instance, whooshing through on a train. The oute does not go through the centre of the city, so you have to change buses if you wanted to travel along it by public transport and, indeed, there are probably parts where there are no buses because the busues, naturally enough, are going to and from the centre.

The point, of course, is that if you lived at one end, you'd have no particular reason to ever go to the other end, but to the person just driving through Preston on their way from somewhere else to somewhere else, it all looks the same. That's the same in the way that Malaga might look the same with endless kilometres of literal totalitarian apartment blocks. It's fractal. As I have said before, each bit of the city, like the Mandelbrot set, is different, but statistically it is, in some sense, all the same.

There are, of course, lots of interesting things that you would walk past if you walked this route and some of them would have more general significance like Deepdale or the Moor Park Observatory. There is a lot of personal significance for me. Parts of the route have been my stomping ground for five decades and I have driven along the whole route hundreds of time, although not, of course, at once. There just used to be a newsagent's at Lane End's that had a sign that ran down the wall of the building facing the oncoming traffic that said "Your Evening Gazette, sir". But since the Evening Gazette is the Blackpool newspaper why would there ne a sign advertising it in Preston? Could you at one time buy the Gazette in Preston? Was there competition between the titles in Preston itself once? Is Robbie's still there? It was a sweet shop and not a newsagents and certainly existed for many years have Robbie himself retired, but I remember it when it was a proper sweet shop in the early 1970s. 

But that something went years ago. Everyone has there own memories. Everything has heft in childhood and I spent mine in one place, so the place is drenched with them for me. Like the fact that you could buy winemaking kits in the Boots in the mid-1980s. "That used to a... That used to be... That used to be different... I remember..." And, yes, there will certainly be people around who remember Robbie's. I remember as a kid thinking there could be people living in these houses who like the things I like. And I eventually met some RPG and sf fans in Preston But even know I drive past a house and think there could be something interesting in there. And there might be, but there probably isn't.

There are all those people, leading all those lives. With all their own histories and all of their own stories. All that sheer weight of stuff. How could one begin to connect to them? Of course, one could just write one of those psychogeographical books about the route.But who reads those kinds of books expect for people like me?    

Little Nostalgia

Hat-tip to the Weasel for pointing out a recent New Yorker article on Stewart Brand and the 50th anniversary of the Whole Earth Catalog. It is, as he says, very New Yorker. Consider this sentence from the author's (Anna Wiener's) conclusion:

I pictured woolly mammoths roaming the charred landscape of Northern California and future archeologists discovering the remains of the ten-thousand-year clock in a swamp of nuclear waste. 

Although one sort of knows what she is getting at, it is unlikely that either 10,000-year clock will be in a swamp of nuclear waste at any point in the near future. The original site is in Nevada and a clock is being constructed in a mountain at Jeff Bezos's ranch in West Texas. Of course, you might tend to store nucleasr waste in remote, geologically stable areas,but, by definition, a mountain is unlikely to become a swamp. Of course, we tend to worry more about places becoming deserts because of climate breakdown, but it is also going to rain more, so some places will become swamps hat aren't presently. Of course, you could take this into account in deciding the location for your long-term nuclear waste depository. Indeed, where to put your depository is exactly the kind of issue that one woulkd expect the Long Now Foundation and Stewart Brand to interested in. It's unclear now paid progress on the 10,000 -year clock in Texas. The last update appears to be from 02011 (as they would have it). The Seminars About Long-Term Thinking though are very much current. Niall Ferguson is speaking on Monday (indeed, Niall Ferguson) and there is a schedule through April 02019. Bet there is something on nuclear waste in there (and Brand's book in which he promotes nuclear power).   

Wiener reports that Brand told her that he has "little nostalgia" for “Whole Earth.”

‘The Whole Earth Catalog’ is well and truly obsolete and extinct,” he said. “There’s this sort of abiding interest in it, or what it was involved in, back in the day, and so the reunion was a way for the perpetrators to get together and have a drink and piss on the grave.

We might feel that's all very well for Brand, but the only thing we have to look forward to is the past. In fact, we have double nostalgia, nostalgia for The Whole Earth Catalog and nostalgia for our first Whole Earth Catalog as the Weasel might have it. I remember my first encounter with Co-Evolution Quarterly, via News That Stayed News- 1974-1984: Ten Years of Coevolution Quarterly, which the Weasel had a copy of and I saw at his house in 1987. I latter got my own copy via Amazon. I once found a bunch of copies of Coevolution Quarterly in a shop in the Cut.   Of course, you can read them online now (if you could can get Flash to work). (The legendary, last, lost issue of Whole Earth on the Singularity is shown, but you not the articles.)

The Whole Earth Catalog, Coevolution Quarterly, The Media Lab, The WELL, How Buildings Learn. There is a lot to be nostalgic about there, both personally and vicariously. As we have said, what is interesting is whatever Brand is thinking. We have an idea of Brand is still thinking and that is certainly interesting (the Long Now material, history). But, as I have asked before, hat is the new thing? I am 50. I hope that 50 is the New 35. So what is that is happening now that I should look into? I know the dull, but important answers of Trump, Brexit, the Alt-Right, climate breakdown, Facebook. But are there other things, maybe even more positive things, that I haven't thought of yet?