The following extract from Permutation City by Greg Egan covers several topics of interest to Samizdatistas and the commentariat. The “Copies” are fully conscious computer simulations of people who have had their brains scanned. The first speaker, Durham, is a biological human trying to persuade the Copy, Thomas, that in the long term he is in danger of being switched off, even though the computer he runs on is private property, by governments claiming the moral high ground.
‘…The privileged class of Copies will grow larger, more powerful — and more threatening to the vast majority of people, who still won’t be able to join them. The costs will come down, but not drastically – just enough to meet some of the explosion in demand from the executive class, once they throw off their qualms, en masse. Even in secular Europe, there’s a deeply ingrained prejudice that says dying is the responsible, the moral thing to do. There’s a Death Ethic – and the first substantial segment of the population abandoning it will trigger a huge backlash. A small enough elite of giga-rich Copies is accepted as a freak show; tycoons can get away with anything, they’re not expected to act like ordinary people. But just wait until the numbers go up by a factor of ten.’
Thomas had heard it all before. ‘We may be unpopular for a while. I can live with that. But you know, even now we’re vilified far less than people who strive for organic hyper-longevity — transplants, cellular rejuvenation, whatever — because at least we’re no longer pushing up the cost of health care, competing for the use of overburdened medical facilities. Nor are we consuming natural resources at anything like the rate we did when we were alive. If the technology improves sufficiently, the environmental impact of the wealthiest Copy could end up being less than that of the most ascetic living human. Who’ll have the high moral ground then? We’ll be the most ecologically sound people on the planet.’
Durham smiled. The puppet. ‘Sure — and it could lead to some nice ironies if it ever came true. But even low environmental impact might not seem so saintly, when the same computing power could be used to save tens of thousands of lives through weather control.’
‘Operation Butterfly has inconvenienced some of my fellow Copies very slightly. And myself not at all.’
‘Operation Butterfly is only the beginning. Crisis management, for a tiny part of the planet. Imagine how much computing power it would take to render sub-Saharan Africa free from drought.’
‘Why should I imagine that, when the most modest schemes are still unproven? And even if weather control turns out to be viable, more supercomputers can always be built. It doesn’t have to be a matter of Copies versus flood victims.’
‘There’s a limited supply of computing power right now, isn’t there? Of course it will grow – but the demand, from Copies, and for weather control, is almost certain to grow faster. Long before we get to your deathless utopia, we’ll hit a bottleneck — and I believe that will bring on a time when Copies are declared illegal. Worldwide. If they’ve been granted human rights, those rights will be taken away. Trusts and foundations will have their assets confiscated. Supercomputers will be heavily policed. Scanners – and scan files – will be destroyed. It may be forty years before any of this happens – or it may be sooner. Either way, you need to be prepared.’
Is simulating a human consciousness P or NP? If the former, then since things like climate modelling are NP, there’s no tension. At present, we are well into the petascale regime, with exascale on the near-term horizon. Computing power is one of the least scarce resources on the planet, as should be obvious from Moore’s Law (if it were becoming scarce, its cose would increase with time; this is, to put it mildly, not the case.)
No idea about your first question. I can think of ways the two are similar (bi-directional interactions between nodes) and different (brains are likely to have repeating structures that can be simplified).
I suppose Egan is positing that Moore’s law has ended. It might happen, and the consequences would be bad, but I think it’s unlikely for a while because physics allows for pretty dense computation and humans are clever.
But the more general point is the way governments like to take control of resources for “moral reasons”. Egan’s example is extreme because private property (ownership of computers and the energy to run them) is never more important than when your very existence relies on the computing substrate you own.
David Gillie writes:
On this occasion, I have a somewhat different view to David’s, though the cause and extent of the disagreement is not totally clear. Still, ever onwards …
Despite wonderful progress on power consumption of High Performance Computers (HPCs, which are named for computational power – ie CPU performance – and not electrical power – ie power consumption), the human brain currently comes in at least 3 orders of magnitude better.
Though this requires some approximation between MFLOPs (millions of floating point operations per second) and MBOPs (millions of brain operations per second), there is no doubt that the human brain requires some 25 to 30 Watts. Such comparably powerful computers as we have (and are they really comparable, or just very useful tools) require MW (that is millions of Watts). I go currently for 1 MFLOP == 1 MBOP, imperfect though that is.
Over the last few years, the power efficiency of computers (usually now measured in Watts per MFLOP) has become an important issue. Useful information can be found here: http://en.wikipedia.org/wiki/High-performance_computing . There is also particularly useful information in the 53-slide presentation of Prof Dr Hans Werner Meuer, a world-renowned expert on the history and progress of HPC: http://www.isc-events.com/slides/london/Supercomputers_LKML_London_HWM_20120418.pdf .
I had the pleasure of attending Prof Dr Hans Werner Meuer’s lecture at the House of Lords on 18th April 2012. The most pertinent slides are numbers 19 and 45, though the latter should be treated with caution as the most power-efficient computers are way down in total MFLOP performance. In addition, slides 27 (repeated at 47), 49 and 50 are particularly interesting from a historical and futuristic predictive basis.
Anyway, back to Rob’s posting and the book by Greg Egan …
If you have trouble with an artificial intelligence: like in the Barclays Bank advert of not so long ago, you could always pull the plug. Also, noting that the true power of the computation is not local, you could also break off the antenna or (more subtly) jam its radio signals.
Meanwhile: Diu hominum. Or so the computer tells me, as I did not do Latin.
Best regards
Yet again, my stunningly erudite observations are with Smite Control.
May the force (of lightning speed) be with them.
Best regards
Nigel, MFLOPS/W is rising dramatically. The new IBM Sequoia machine which just took the lead in the TOP500 comes in at about 2GFLOPS/W. The slope on the plot of the fastest TOP500 machine works out to a an e-folding time of about 20 months (i.e. doubling about every 14). Absolute power consumption seems to be roughly static. I don’t think the speed of brains and computers are commensurable, but I can’t see why a biological substrate should always be more efficient, and many reasons why it shouldn’t. Three orders of magnitude growth in a computing metric is a very short time.
If Greg Egan’s the guy I think he is ( I gave away my v.extensive SF* collection before fleeing the ongoing UK nightmare) he’s the guy writes about AI & sets whole novels in computer simulated realities.) then in the discussion your having your rather missing the point. It’s hard enough predict what the state of tech will be next year, let alone in a decade. I’d imagine what Egan’s trying to do is explore how a society would handle members of it migrating to an existence as software. What happens when the rolling over of the generations comes to an end & rich auntie isn’t going to die & leave you her wad but continue to be around, if only as a voice on the phone or an electronic image. Which is all you’ve seen of her for the past ten years so what exactly is your problem? It was Egan made me realise the whole- ‘self concious AI goes rogue & fills the universe with robots’- SF plot is anthropomorphic bollocks. An AI’s only interest would be increasing it’s clock speed so it can think quicker. The route to do that, as we know, is to get smaller because the limiting factor is the speed of light. As it gets smaller & faster it loses interest in the macro universe. What it can see of it is only the input of its sensory array & there’s not much happening out there. As computational speed increases the universe slows down & stops. Eventually it’s manipulating quarks, or something even finer grained in the quantum world. But long before that it’s disappeared from human view altogether. Maybe it slings us an annihilating couple of megatons of antimatter as a parting shot. Maybe it doesn’t. It certainly isn’t interested in taking over the galaxy when it can live through several eternities a second.
*Sorry Saizdata folk, but SF to me isn’t “science fiction”. Maybe speculative fiction or speculative fantasy but not Star Trek. I gave up on Dick & the rest of them in my teens. A galaxy’s worth of small town America or cowboy’s & indians in spaceships. Never even read Burroughs. Niven appealed because his aliens were actually alien. Gibson I like because we’re now living in the world he wrote in the 80s. Remember going to a computer fair off the Tottenham Court Road & buying some RAM off a Russian & a drive off an African in robes. The Sprawl subsumed London about 2004. Read his latest stuff & realised I actually know ‘cool hunters’. OK, some of his tech happened differently but his social stuff was prescient.
Social stuff.
I’ll tell you how I spent a couple of days recently. Spent them with a woman I met on the net. Not off a dating site. It was someone got in touch through a mutual – one removed. Needed some assistance. Net’s like that, isn’t it?. The dialogue strayed from the original topic, as it does. We’ve been chatting on & off, using various means for a while. Probably know more about each other than some couples because there’s things you can discuss, with a stranger’ll never cross paths with anyone in your world, that you wouldn’t with your closest friend. So there’s two people find they’re simpatico & fancy taking it to the next stage. Agreed to meet in a country convenient to both, clicked & spent the next 48 fucking their brains out. Facilitated by easy travel, credit cards, e-mail, Skype…..And….And……Neither of us speak more than half a dozen words in the other’s language. Never have. I can’t even pronounce Slav languages. She’s limited to hers & a bit of Russian from school. We dined in restaurants, & sat up in bed sharing a laptop plugged into mobile internet. Connected to Google Translate. Before we parted we were seriously wondering if we couldn’t make a go of it. If we wanted. Conquer the language thing as & when. Or would we have to? Depends on tech. How far is voice recognition & simultaneous colloquial translation away? Couple years? Smartfones, The Cloud, Bluetooth, Wifi are now.
Ever read a story with a plot line like that? I haven’t. Wouldn’t have thought it probable if I had. But I’m living it.
That’s what I see SF’s for. It’s not to predict how much computing power you’d need to upload auntie.It’s how you’d interact with an uploaded auntie. Coz you just might have to.
.
A lot of Egan’s books deal with what consciousness will mean in a cyber age. The trouble is- he concentrates too much on the future, and not enough on the characters. I find it hard to empathise with them, so I don’t get involved in the story.
Bloke in Spain: you are right. That is the Egan you are thinking of and yes, that’s the point. And I agree with you, except I imagine that despite being able to live many lifetimes per second, AI might still want to expand and turn all the matter in the universe into computronium, eventually. I don’t mind so much about that; I’d quite like to join them.
Nuke: oddly enough that’s why I like Egan. He is untroubled by criticisms about characterisation and all the usual “literature” conventions. He concentrates on what he does best, which is ideas compatible with physics.
Incandescence is another of his. It’s my perfect novel. I should review it. In answering its critics, Egan wrote:
Judging from his footnote, bloke from spain would agree.
“AI might still want to expand and turn all the matter in the universe into computronium, eventually.”
Not so sure about that. Unless it can find away to propagate information faster than light. Or it’s got an inbuilt drive to replicate, which is very much a meat fetish to borrow a Gibson derogatory. More likely it’d want to turn the whole of a single atom into computium. A universe wide AI, at light speeds, wouldn’t be ‘thinking’ at all in any meaningful sense. Its ‘neurons’ would be expanding, as the universe itself expands, as fast as the information they were carrying. Even at galactic dimensions, its clock speed would be in the thousands of years per operation.
I’m pretty sure I read Incandescence although anyone who’s read Egan would understand a reluctance to discuss the details at a distance of several years. The guy’s stuff is at the ragged edge of my limited comprehension. But yes, the ‘science’ in a lot of SF is not discernible from magic because that’s all its authors can cope with. And for someone who enjoys writing, much of it’s laziness & cheating. Get your character in an impossible position & dream up some bit of claptrap to extricate him. Never could understand the attraction of Asimov’s Foundation novels. Or most of Asimov for that matter. Adventure yarns given a coat of metallic gloss.
Gibson fascinates me because with each round of trilogies he’s got closer to the present. His latest is set on an alien planet, a mystery to us. Ours. Now.
And he appreciates the impact of tech is most pronounced on those at the bottom of the pile. The internet, e-mail, mobile phones. How much difference have they made to the rich & powerful? Marginally. They already had good communications. The world was already a small place for them. Putting a mobile in the hands of the dealer on the street corner. The Thai hooker who can Skype her mother, half a world away, between tricks & transfer the money home with a mouse click. The kid from the Rio favella who’s as much net presence as an Eton scholar or a teen in a bombed out building in Syria.
And Egan’s made me resent not living in a hypertexted world. There’s no rollovers I can click on to explain it. I’m in increasing symbiosis with tech. I see the world through the prism of GPS. 5 hundred km in fewer hours, to hit a point in a strange city with 2 metre accuracy. Knock at a door already recognised from Street View. Without net access, reduced to the mere 6 languages loaded on the laptop, I’m almost deaf & dumb.
F**k knows where we’re headed but SF’s our only guide
I cannot help but wonder as to what happens to a copy of the human brain from an IP point of view. Suppose, for example, that I can copy my brain – can I see version of it, and if so, could I claim IP on this, given that my personality is involved?
Must be a SF plotline in there somewhere!
Similar plots have already been done. James hogan had a story about robots on Titan, who were carrying downloaded memories of an alien species within their memory-banks, even as they went about their ‘lives’, blissfully unaware. James even worked out the rudiments of a machine ‘ecology’.
And Egan had a story about kidnappers taking the stored downloads of people and holding them for ransom, in an age where downloads are legally people. He has some good ideas- if he had more likeable characters, he might even rival Peter Hamilton, who deals with similar ideas in his ‘Commonwealth’ books.
I am now not sure where you are getting your information,
but good topic. I must spend some time finding out much
more or working out more. Thank you for wonderful info I used to be on the
lookout for this information for my mission.