Friday 4 July 2008, 12:42 PM
Do the maths: HAL 9000 by 2018?
There's a new science in town: connectomics. A specialised form of neuroanatomy, connectomics is in the business of mapping out the brain's networks in particular, how the various functional modules already identified connect to each other. While this has always been of interest to brain researchers, it's only recently that the IT's been good enough to embark on seriously detailed mapping and fascinating new results are already turning up.
For example, work at Indiana University, University of Lausanne, Switzerland, Ecole Polytechnique Fιdιrale de Lausanne, Switzerland, and Harvard Medical School has uncovered a "superhub", a structure between the two halves of the brain where many networks converge, and that's active all the time. Exactly what it's doing and what happens when it goes wrong are two very interesting questions.
Connectomics has a higher aim, to produce a complete map of the brain's networks akin to the sequenced information from the human genome projects. That's ambitious, but by combining new extremely high resolution scanning technologies with computational analysis that pulls structures out of the data it seems entirely realisable.
Meanwhile, other intensive data crunching is cogitating away with IBM's Blue Brain project, where the team has built "a rat-scale cortical model (55 million neurons, 442 billion synapses) in 8TB memory of a 32,768-processor BlueGene/L" That's a long way from the ten billion or so neurons of the human brain 180 times smaller, in fact but in Moore's Law terms, that's around ten years. And the researchers are most definitely looking ahead: "Our long-term goals are to develop novel brain-like cortex size computing architectures along with appropriate programming paradigms, and to evolve C2 into a cortex-like universal computational platform that integrates and operationalizes existing quantitative neuroscientific data to build simulation of large networks of spiking neurons and a powerful learning machine: a cognitive computer" (from their report Anatomy of a Cortical Simulator - pdf).
You know what that means. Doesn't matter if you don't - it will know..
A lot of that work is going on in the same places as the connectomics research so no prizes for guessing where that's all heading. We may have missed the starting gun, but the race to produce a full working model of the human brain has most certainly started.
Wednesday 2 July 2008, 2:46 AM
When blogs collide...
Out in the blogosphere, a most intriguing story is well, not unfolding, exactly. That's precisely what it's not doing. It's more like a well-written psychological thriller, where event piles on event but the truth becomes ever more difficult to untangle.
It all revolves around Boing Boing, one of the Web's longest-running alternative publications. Subtitled 'A Collection Of Wonderful Things', it's run by a small coterie of high profile individuals and is dedicated to a wry, enthusiastic take on events and inventions best categorised as youthful, creative, energetic eccentricity happening on the edges of the mainstream. As befits such a nexus, it has a strong cultural disinclination for the censorious, the corporate (unless it happens to be Disneyland) and the incurious: it is perhaps too easy for the cynical to read Boing Boing's passion and self-confidence in such matters as smug self-righteousness, and its fondness for reporting on the extramural endeavours of its team as irritating self-promotion. Me, I like the place.
Recently how recently remains under discussion Boing Boing decided to remove all posts related to one Violet Blue, a friend of the blog whose field of expertise was and is "sex culture commentary, accurate sex information, erotica and more". This sudden excising attracted the attention of many, including those on the Metafilter group blog, another very enjoyable site whose remit overlaps that of Boing Boing but which is lower profile and driven by a much broader group of contributors. Many of those, it transpires, find Boing Boing extremely annoying the Metafilter thread on the Violet Blue affair is now longer than that provoked by 9/11. The schadenfreude they feel at such a high-profile defender of online cultural freedom behaving in such an unusual fashion has been described as that inspired by finding the leader of an animal rights campaign being found in a restaurant tucking into a large and bloody steak.
This reaction has only been heightened by Boing Boing's extreme reluctance to explain why the removal happened. Such official response as has been forthcoming has been defensive, lacking in detail or explanation and managing only to further inflame the meltdown. Third party comments have been quietly deleted on Boing Boing and on other, related blogs and those quiet deletions have led to further rooftop shouting elsewhere. Violet Blue herself affects a naοve bemusement.
There are, of course, any number of theories. Is it a lovers' tiff? A case of wrongdoing, badly handled through a wish to avoid too much worldwideweb mudslinging? A massive egofight? Or are legal machinations at work behind the scenes, of the sort precipitated by expensive lawyers slapping on injunctions? For what it's worth, which is precisely the value of the paper on which this is written, my money's on the latter. You can taste the frustration in some postings by the protagonists, who want to say more but are utterly unable to do so.
Certainly, there's a lot at stake Boing Boing's extremely popular and at the heart of a considerable commercial empire, and as such seen as a leading example of new-wave Web publishing done right. Just as certainly, there's a lot of passion involved. I find it rather heartening; the legacy media is no stranger to similar secret battles played out in public, but in a code only discernible to the cognoscenti and the readers of Private Eye. It's to the credit of the blogosphere that such peculiar and distortive happenings are far harder to hide from readers.
What happens next well, we've all got ringside seats. The truth will come out, I'm sure, and I hope that the collateral damage isn't too painful to those involved. Meanwhile, I commend the case to anyone interested in the anthropology of that new yet very familiar breed, Homo cyberneticus, into which we all appear to be evolving.
Tuesday 1 July 2008, 4:23 PM
Microsoft's new licensing may not fly
Microsoft's new Select Plus licensing scheme is an intriguing modification to an increasingly problematic arrangement: the exchange of large amounts of enterprise money for the right to use software you've been paying for for years.
Setting a licensing scheme is like flying a plane. Too low a price for a fixed volume Windows has saturated the market and you lose money without gaining anything. That's like reducing your airspeed to a dangerously low level. Too high a price, and you drive users away: although Windows has a near-monopoly on the enterprise desktop, the environment around it is evolving rapidly and there are many more options this year than last. That's where your wings drop off.
Microsoft's great worry must be what pilots call coffin corner a particular combination of speed and altitude where you can't speed up or slow down without dropping out of the sky. Too fast, and turbulence disrupts airflow over the wings, causing a catastrophic loss of lift. The nose goes down, control surfaces stop working and you're faced with an engaging set of problems. Too slow, and you stall: nose goes up, tail goes down, control surfaces stop working and you slide bottom-first towards the ground. Again, you're likely to spill your in-flight refreshment without having to worry about the dry cleaning bill afterwards.
So it is with Windows licensing. The one question Microsoft doesn't want anyone asking is What are we getting for our money?. The company is quite open about this the Select Plus deal is designed around what's known as inertia marketing: sign someone up for a repeating deal and then hope they forget about it. It works in publishing and telecoms; why not for enterprise licensing? Microsoft even list this as a user benefit - Perpetual agreement. Keep renewals at the IT budget level rather than the boardroom level.
Inertia marketing rarely benefits the customer, and this is no exception. Microsoft benefits by having a constant fixed income that's high enough to keep the company in Xbox development funds but low enough that nobody spending the money will ask questions. The enterprise IT department's job in all this, thinks Microsoft, is to conspire in not rocking the boat. In cash-strapped times, this is not how any enterprise IT head is likely to think.
Don't take Microsoft's word for it. Plug the Select Plus into your spreadsheets and run through a few years then ask whether you're getting the best deal in exchange for that bottom line. Then try some other numbers for some other ways of working, and see how happy you are to keep the Redmond Express in the air. A little turbulence may be good for their souls.
Wednesday 25 June 2008, 9:40 PM
Back from Zurich...
... and I think I'm recovering. There is something very odd about a place where I can take in an IBM press conference - IBM, for heaven's sake - where the presenter says "This is a nanobot we're going to INJECT INTO YOUR EYEBALL and GUIDE IT WITH INVISIBLE POWER RAYS and let it SCRAPE YOUR RETINA. Okthxbi.", and I just sit there and think "Oh, that's cool.". It's only now that I've returned to Holloway that I begin to see quite how strange it all is.
Once upon a time, I went to IBM press conferences and they buried me beneath acronyms about storage and networking, and I learned very little. I'm still trying to imagine what it would be like going to a MIcrosoft press conference where Ballmer described a nanobot whose job it was to corrode me from within, and I thought it a good idea.
Ah, Switzerland. A country where you can't get a sandwich after ten thirty, but you can buy cannabis tea from vending machines in the main railway station. Perhaps there's a reason that nanotechnology, CERN and the World Wide Web all come from the place...
Wednesday 25 June 2008, 1:27 AM
Nanotechnology - small wonder it's coming
OK end of play, following a morning at the IBM Zurich labs and an afternoon at ETH Zurich, one of Switzerland's leading universities. It counted Einstein, Roentgen and Pauli among its physicists, and regularly adds Nobels to its trophy cabinet. It's that sort of place.
I've seen some great things which will need a bit of beard stroking and research to properly regurgitate, but here are some immediate thoughts.
Nanotech is coming and it's going away. It's coming because it solves real problems and makes good use of all the expensive lessons we've learned refining semiconductor physics and production, and it's going away as a concept because it's going to be part of everything. There is nowhere else for chips to go: the introduction of the 80386 is further behind us than we are away from all the roadblocks at the end of classical semiconductor development. And nanotech is going to become a huge part of the future of chemistry, biology and physics: nothing else gives us the power to work at the scale that really matters to us every system that makes us up can be considered as nanotech, and wherever you look in energy, environment, food and materials science, developments at the most intimate level have the biggest potential impact.
Before that happens, here's where it'll turn up first. Medicine and health. You name it molecular analysis of samples, micro-surgery, drug production, monitoring implants, all are huge markets waiting for the increase in efficiencies, better procedures and plain old cost savings that'll happen when we better engineer tiny things that interact with our bodies. For example: tiny robots small enough to fit in a particular size of syringe that can be powered, controlled and monitored from outside, and which do real surgery on retinas. Why that size of syringe? That's the largest that can inject into the eye without requiring sutures.
As for other aspects: when you've seen nano-sized helices busy propelling themselves through a liquid like so many bacteria, heard researchers talking about the practical challenges of energy harvesting (you can't store power in things that small, so they have to take it from their environment), and seen 500 nm polystyrene beads precisely positioned on the surface of 100 micrometer glass balls that have themselves been precisely positioned, you don't doubt that the basic toolset of nanotechnology is advancing as quickly as people can think the ideas up.
IT well, it'll be slower arriving than you'd like but probably as quick as it needs to be. Spintronics, where you make things happen by the way electrons behave rather than by moving them around, remains tremendously exciting and still largely unknown. We have it already in hard disks; it's moving into solid state memory (where yields are still a huge problem, but that's being tackled) but is still an unknown period away from making useful logic. There are plenty of questions of basic physics to solve before that happens fortunately, they include plenty of discoveries that are exciting but badly understood.
Making exotic 3D transistors that work faster, smaller and with much lower power is going to happen first: in fact, it's already happening. (Making circuits out of them that aren't severely compromised by cosmic rays is, however, still an interesting problem). Likewise, clever optical interconnects are high on the agenda. These are two basic new technologies that will be around when the old ideas we've relied on are out of steam.
Tomorrow is a big announcement; we've been told what it'll be but sworn to secrecy until it happens. It's not a breakthrough or a big discovery, but it is important and will be another pointer towards what happens next.
See you then. Keep it tiny.
Tuesday 24 June 2008, 12:41 PM
Nanowires!
Just in the lunch break (yes! Fed at last!) at the IBM Zurich labs show-and-tell day, which as usual is full of sessions that overrun and no time to do anything except scribble the most incomprehensible of notes.
One of the more interesting things happening here is nanowire the creation of very thin whiskers of material that can be used to make novel and potentially groundbreaking semiconductor devices. So far, these have mostly been made out of silicon, and the way you create a forest of these is disarmingly simple.
First, take your silicon wafer. Then, print a pattern of nanometric gold dots on it I'll get back to how IBM do that later. Take your wafer covered in tiny gold particles, heat it up until the gold melts and stick it in an atmosphere of a gas containing more silicon. The vaporous silicon dissolves in the gold dots until no more can be absorbed a supersaturated solution at which point, the silicon in the gold starts to crystallise out at the barrier where the gold is resting on the silicon wafer. This creates a small bulge in the wafer, pushing the gold dot upwards. The process continues, with the bulge turning into a whisker, then a wire, all the time growing upwards.
This isn't fast only around 50 nanometres a minute but you can do billions at a time. Once you have your nanowires, you can deposit other stuff around them to create transistors that work at extremely small voltage swings, extremely fast. And it looks as if this will work down to around 9nm, which will be just in time: from having devices looking good in the lab to getting into development normally takes five to six years, and then it's another five to six years from there into production. 2020 is a reasonable guess.
Just about any material can be grown as a nanowire indeed, by changing the gas composition during the process, you can change material and grow different zones and mixes of elements. IBM is particularly interested in what are known as III-V elements, which cluster around the part of the periodic table that's good at making electronics, because they're also optically active: we haven't had the optical interconnect talk yet but IBM, like everyone else, is absolutely convinced that there's only one way forward for high performance computing, and that's by using more and more light.
Right, off to the labs...
Tuesday 24 June 2008, 12:35 AM
No cheese in Switzerland
It's half past one in the morning, Continental time, and I'm about to go to bed without any supper. I'm over in Zurich courtesy of IBM, which is going to be showing off a lot of its new nanotechnology developments tomorrow and has a big announcement presumably about something very small on Wednesday morning.
The supper, alas, is a victim of old technology and the peculiar Swiss disinclination to take your money after sundown. I flew from London City Airport, normally one of my favourites you can go from arriving at the airport, through check-in and security and onto the aircraft in ten minutes, if you time it right. Not this time: even though I turned up with slightly more sane margins for error, the ticketing and check-in systems were down.
And stayed down for at least an hour, during which time the on-site techy (helpfully displaying the name of his company, ESP,specialising in keeping critical IT systems up and running 24/7 around the globe for airlines, airports, the travel industry and corporate enterprise according to its website, on the back of his hi-vis jacket. A pal is hassling their PR even now) ran around a lot and avoided answering questions. In the end, the check-in crew was reduced to doing it manually, a task which they clearly hadn't practised and clearly didn't work. Nobody checked my e-ticket, and the hand-written boarding pass wasn't acceptable to security without an escort. There was a lot of running around, and a lot of very unhappy punters and, in the end, an extremely delayed flight.
Which meant I got into the hotel at half past eleven, to find that even room service shut down an hour earlier. You're supposed to keep a certain hunger about you if you're going to be a good journalist. That won't be a problem...
Thursday 19 June 2008, 8:03 AM
Intel's floating bodies
Intel got on the phone on Monday to tell us chip-centric hacks about a bunch of papers the company was presenting to the 2008 Symposium on VLSI Technology in Hawaii. I've been trying to turn the notes from that briefing into a full news story, but on mature consideration it's more suitable to the informality of a blog post - when a company braindumps unconnected bits of technology without a real common theme, it doesn't fit well into the traditional channels.
First and most interesting is a new memory architecture - floating body cells, an evolution of dynamic RAM. FBC integrates a number of components from existing DRAM designs into a single specially designed transistor. It uses a special very thin silicon on insulator technology which allows us to use a lower voltage than today, Intel said, it's more than a factor of 30 smaller than current cache memory at 45nm, and ten times better than anything anyone else has published. It's conceivable for 15nm and beyond. - so that's around three years away. It gets some of its size advantage because it's basically a single transistor memory, compared to the six-transistor static cell used in processor cache, but there'll be additional complexity because dynamic memory is more difficult to control than static.
There were more disclosures about Nehalem, In particular, the first version of this new architecture will have four cores all with independent control of their voltage and clocks - and the ability to change those values very quickly and to a high degree of precision. In fact, says Intel, the cores can reconfigure every clock cycle - and since they're equipped with very good ways of checking their own performance, they can operate in the most efficient way possible without having to have lots of slack for safety. From an engineering point of view, this is a big step towards an ideal that's been mooted for ages - circuits that have a lot of knowledge about how they're working and a lot of intelligence of how to adjust that on the fly.
Another journalist (missed his name - apologies) brought up the very good question of whether this means that traditional speed ratings and performance metrics for chips are no longer appropriate, and whether Intel would be abandoning them. Intel's answer was intriguing: "We debated this quite a bit. When we talked to partners, people did not want that. A tremendous amount of innovation has gone into avoiding that. Internally, the chip is adapting, but it is deterministic from the outside."
The same theme - of chips deciding for themselves how best to operate in situ - came up over signalling and bus speeds. It's hard, expensive and inefficient to set up external tests during production to check the limits of how fast a bus can operate. Thus, Intel is building more and more tests into the chips themselves - transistors are cheap but performance isn't - to measure key parameters such as clock jitter which directly limit how fast a bus can reliably run.
All rather intriguing. For a long time, there's been a tacit assumption that Moore's Law is about making things smaller, therefore faster: the truth is that it's about making things smaller, full stop. When faster stops happening, smaller has to mean smarter - and there's only so much that companies like Intel can do without the software side of the industry getting smarter too. No more free rides.
More on that particular chain of thought later.
Monday 16 June 2008, 1:32 PM
Twitter - the clue's in the name
Battling with daily info-load issues myself, I have never been much of a fan of the whole Twitter thing. The idea is fine - an extension of the status line update informal communications channel that's been a part of online life since the first time that mainframes and minicomputers allowed their users to see what everyone else was doing. But the practical effects of voluntarily immersing oneself in yet another layer of asynchronous noise (can you have synchronous noise?) did not appeal.
Nonetheless, about a month ago I finally allowed myself to be talked into Twittering. At least, I tried: the site was down.
I tried again, and managed to register. Since then, I've spent more time trying to make bits of it work than I have enjoying the service - and when it does work, it seems to have two modes: silence, and avalanche.
My latest experience has been just as daunting. The woman who talked me into trying it IM'd me asking "Can you get Googletalk working with it?".
No. I followed the online instructions: no good. I searched for enlightenment online, and found plenty of "Do this, and it will just work" recipes. Only it didn't.
Finally, I asked a friend. "IM's been down for weeks," he said. "First thing that goes when they have capacity issues. And they always have capacity issues." He then pointed me at the status.twitter.com page, which indeed listed IM as being offline. Of course, once he'd done that I immediately spotted the link to the status page on the main Twitter input. D'oh.
I have some sympathy for Twitter. Just not that much. Have a big success online, and you're riding the tiger: revenue doesn't scale with popularity, but by gosh your tech costs do. It's also not unknown for all sorts of architectural problems to appear once something gets properly wellied, and for them to take time and money to fix. Often, you'll have features that don't stand up and have to be turned off - or even abandoned. That's part of the game these days, and anyone who tries new services should be aware they're part of the Great Experiment, where nobody knows what's going to happen next.
So, Twitter, turn the darn things off. Remove the "Here's how to sign up to IM" options. Slap "Closed for renovations" over the thing. Don't leave your site as a guessing game. Don't expect people to find your status page - no, not even if you leave the link in plain sight. Consumer electronics manufacturers know that even if the first thing the punter sees when opening the box is a huge piece of paper with "READ ME FIRST!" written on it in 72 point letters of fire, that paper is effectively invisible. That's not what the punters are there for.
It is about managing expectations, and that's all about understanding how people work. Twitter doesn't - and for a service all about people's intimate behaviour, that's a killer flaw.
Previous
