Heh, maybe this is a sign of how bad things already got, but your nightmare scenario doesn't really scare me very much. I mean, I hate ads, and I hate them because they're a nuisance, not because they tempt me. Your post made me picture a future in which the ads that hit my eyeballs are perfectly targeted based on all my communication data. What would this mean? No more "new season of 24"-induced nausea, no more yeast infection cures, erectile disfunction remedies, no more ads for Kevin Federline.
So what ads would I get? That might actually be interesting: Probably ads about recently released physics books, about how Neurosis will be playing in Buffalo (they'd know where I am!), about a comic book convention in Rochester, a sale on my favorite strings at the Guitar Center... - and I'm sitting here wondering whether this would be so bad. After all, these ads would each very likely result in me spending money, greatly increasing the efficiency of the whole advertising system. Potentially, this would mean that I wouldn't need to watch as many ads. Once they collect enough data, they would realize that most of them are a complete waste anyway, and that it would be wiser for them to just not bother with them. Also, this would really lower the economic barrier for submitting an ad to the media. It might be profitable for a local bar to actually advertise their decision to have Wednesday night trivia, because only people who live in the neighborhood, go to bars and like trivia would get the ad.
Even better would be this: They let me select an "I'm not interested in this sort of thing" option along with ads they show. Of course they wouldn't be contractually obligated to respect all my input, but... generally that sort of input would be good for both the consumer and the advertiser.
Agreed, and this is great news, because that sort of GPU will be so CPU-like that it will definitely also be able to do all the other stuff (besides games) that today's computers still do far too slowly: Media encoding and image/video rendering and processing. In all other categories, a bargain computer of today is plenty fast enough so that speedups will be almost imperceptible. I fully expect future 1080p video encoding to be done on super-parallel multicore CPU's whose main role will be to render games. And I expect it to be done very very fast, much faster than realtime even with very complex codecs.
I hate GUIs too, and you can be sure I'll be lobbying for a command-line-only interface for the Gimp. It might have a steep learning curve, but can you imagine how powerful and efficient that would be?
Are you sure the hand charger is standard with the OLPC? CNN mentioned something about a pull string, but the OLPC website says this cryptic thing:
True, early prototypes included a hand crank, but it was removed in subsequent versions. The actual shipping units will use an off-board human-power system, connected to the power brick. Candidates include a foot-pedal charger similar to the Freecharge portable charger, solar panels, a crank, and a pulley system.
This leaves the impression that they haven't even worked out what this manual charging method will be, and might be leaving it for future editions of the OLPC. If this is wrong, can someone post a link that shows the manual charging system?
Thank you for that fine and insightful rant. I clicked on the comments page thinking I'd have to compose something similar, but instead I think I will go grab lunch.
I think this is a good point. After all haven't we always said that ATi drivers (for windows) suck? And those were written by professionals, the best that ATi could hire, and they certainly had access to all the specs. Now, I'm convinced that if all the relevant documentation is released, the open source drivers will be better than the professional ones, but that's because the OSS community will put countless hours of grueling work into the project. It's definitely worth it. Crappy graphics support has been a curse for Linux in many ways. We will all have a good belly laugh when Windows users look at Linux and envy us for our graphics drivers.
(And seriously, won't there be a Windows version of the open source driver? And if so, might it not surpass AMD's own Windows driver? And might this not be a part of AMD's strategy, out of recognition that everyone disses their drivers and that their coders cost them too many salaries? Probably not, but this will certainly do a lot to make people finally reconsider the undying meme that ATi cards are better but their drivers suck.)
You could be right, but Roman is in a tough position, because he's arguing for a change that he thinks is big, and Ingo seems to be trying to sap his enthusiasm by telling him to essentially "work on what we're doing" when Roman wants to have a debate about the best architecture for the scheduler.
In order to help give substance to the debate, Roman coded together some proof-of-concept stuff, but instead of his architectural ideas being looked at seriously and critically, Ingo instructs him to strip away most things and "well use it." That really should seem to everyone on the sidelines like Roman's ideas are being ignored without debate. Now, maybe Ingo is polite, Roman's work just sucks, and Ingo won't confront him on it. But if that's not the case, maybe there should be a (non-flamey) debate about the best architecture for the scheduler.
I think this is a smart move by Nokia. Certainly, there will be head-to-head comparisons of the two phones, but I'm confident that Nokia has enough good engineers do to actually make a good product. If there are reviews plastered on every website to the effect that Nokia's whateverphone is just as good as the iPhone but with faster internet access, better standards support... AND IT'S UNLOCKED...
...then many people who were made to slobber by Steve Jobs will be stampeding to buy a Nokia. Sure, there are people who want stuff from Jobs because they think he's the iMessiah, but most Apple customers buy Apple gear because the marketing convinces them that it's the best stuff available and priced fairly given what it can do. If it becomes clear that the iPhone isn't better than Nokia's knockoff, I don't think there will be many people who will want the inferior or more expensive product just because of its logo.
This means that Nokia can spend far less on advertising, so offer their phone for less and on more reasonable terms. For example, they can let carriers subsidize the price as a reward for signing a contract, so to the customer, the iPhone will look twice as expensive as Nokia's potentially superior knockoff.
It's a brilliant move, and it costs Nokia almost nothing - minimal advertising and no wasted engineering effort, since every major phone company needs to develop a modern touch-screen interface anyway. Might as well catch a free ride on the Apple hype-wave while they're at it!
But do they ever get to exactly the same value? Suppose they start off at.2,.4 and.7 and they get amped up to.6,.7 and.9. You should be able to re-expand these to their original values, right? Does it really ever happen that signals which start at different values get squashed to exactly the same value?
I agree about how it should be done, but I'd be surprised if the original dynamic range was unrecoverable from the modern recording. It's just that the differences in the levels were shrunk. They weren't completely destroyed. That's why I think they could potentially be re-expanded, and my question was: Why is there no common hardware or software that does this?
I don't think that's right. I worked at a radio station and we had our own compressor that our signal got fed through. So did every single other FM station. There is no need at all for the CDs to have this compression on them, since the destruction of the dynamic range is very easy to do for a radio station.
If you do a search for dynamic range expanders, you'll get lots of hits from ebay, because companies like dbx no longer seem to make standalone componenets for re-expanding the dynamic range that was squashed in the studio. And theoretically, the process is reversible, right? I mean, as long as there was no clipping but only squashing, a good processor should be able to undo it. In fact, that's how "high-end" radio used to work long ago: the station broadcasts a squashed dynamic range but then the radio at home re-expands it restore the natural range.
So now my question is this: Why the hell are we not doing that now, with this music-squashing epidemic in full force? The old arguments for why we should squash the dynamic range no longer hold. One of them was to increase the average volume compared to the hiss. Well, we can control hiss better than ever. Another reason is that amplifiers that play music which is on average quieter need to be turned up louder, which adds noise and distortion. But amplifier circuits are much better now, and this effect is tiny.
It just can't be that hard, from a caclulation point of view, to re-expand the dynamic range in a squashed recording. Real-time sotwtware should be able to do this easily. Yet I don't know of any such plugin. Google revealed that a shareware plugin called DFX ($20) might do something like this, but the details are unclear. Am I missing something? What's holding back the dynamic range re-expanders that so many of us want?
Wow, that was some misleading writing, though some very interesting research. I wonder if there's a practical way to observe actual plasma on this level, to see whether the simulation mirrors actual plasma physics. Also I wonder if there is an upper limit on the size, complexity and longetivity of plasma structures. It's hard to imagine something that hot would be very stable, though I'm prepared to be surprised.
I'm pretty skeptical though. If evolving structures are so common that we see them even in a low-powered simulation, and every single star has so much freaking plasma, where are our plasma overlords? Or maybe that's hell, and those structures are just... the souls of the damned! Oooh!
I thought of the same article. We need an explanation for why genetic diversity decreased so much in Europe and not in Africa (for example). And the mechanism of the wealthy families replacing the poor is a very plausible suggestion, since it has been independently shown to occur in Europe and not elsewhere. Of course the plague had something to do with it too, but that's not enough to explain the whole effect. It's important that the same population pressure applied for more than 30 generations, and that's long enough for some pretty impressive consequences to emerge.
The whole point of the industrial revolution was to reduce the proportion of people needded to work the land. It's not just about harvesting. Ever tried hand-thrashing grain with a flail? Neither have I, but it looks pretty labor intensive. Much harder than firing up a steam-powered thrasher. Sure it takes work to make one, an entire village can use it. So if you want to provide an answer for why these things were not made earlier, the answer cannot be that there weren't enough people for it, as though industrializing takes extra work. Industrializing saves work. And the process is continuing... and will pick up as AI and expert systems get better.
It's not quite that "all the lazy and stupid people died out" - it's more that the families disposed to working long hours and delaying gratification were the ones that, in each of 30 generations, had a survival advantage. And 30 generations of the same selection pressure really could make a lasting difference. But what the argument needs is evidence that in places like China, the same population pressures did not apply. Otherwise, he's not offering an explanation of why the industrial revolution was spawned in Europe and Asia didn't have its own home-grown version, despite a favorable climate and plenty of coal.
It's silly of someone so smart to claim that technology has been driving social change since the mid 18th century, because it was it was less than a century later that Marx put forward the view that technology is the only driver of ideology and social change ever. He didn't call it "technology," he called it "means of production," but we recognize what it is. Seemed pretty radical to some people then; funny he now seems so right.
Of course, Marx was different in this way: He did make one prediction about the future whose means of production were unknown to him: he thought there would be a people's revolution in which people would take control of the technology developed in the capitalist era, because of the inevitable resort to artificial scarcity that the capitalist system will increasingly have to turn to. Scarcity will need to be artificial because technology will be able to meet all the basic and many of the advanced needs of everyone in the world. Capitalism doesn't work in situations of plentitude, so there is no market for breathable air (yet). So the artificial scarcity that Capitalists will need to create will eventually get so ridiculous that people will just depose them. As far as futurism goes, I think this outline is aging rather well.
And by the way, this is much closer to what Marx actually said than what most "Communists" claim he said. The Marx I read never advocated a revolution, resource distibution, or any of that other socialist stuff. He was a dialectician who thought that history has an inner logic and moves forward inevitably. Pleading with people doesn't move history; technology moves history. He argued pretty forcefully that Capitalism isn't the final system, but not because he was trying to stir up a revolution. It was just to convince people that it can't last, that, like every earlier technological/ideological era, it will be undone by the tools it eventually creates. So if Capitalism creates automatic strawberry harvesters because Mexicans get too expensive, and intelligent robots and fusion powerplants and workerless factories, it will eventually make the gear of it's own demise. Marx repeatedly extolled Capitalism for being so damn good at producing new technology in the most efficient way possible. It was Lenin, not Marx, who thought that a society can leap past all the stages of industrial and post-industrial capitalism and start a revolution with just an ideological vanguard. Obviously, that didn't work out. Marx was clear that technology drives ideology and not the other way around.
Maybe technological civilizations are intrinsically self-steralizing by war, but why this should be a universal law is just as mysterious as the Fermi paradox itself. Certainly it seems possible that a civilization (even we) could disperse widely enough and soon enough so that no single war could wipe out every colony.
As far as consciousness retreating into virtual worlds, I find this rather probable. However, if civilizations in our galaxy have done this, we would notice it. Surely, some of the simulations would require increasing resources to maintain, because simulations get better the larger and more detailed they are. Give it a few years of increasing energy demands (say a million) and you should see solar-system-scale engineering projects, and then projects that go beyond their original solar system.
I'm trying not to make too many assumptions about life in general, but one safe one is that when habitat is available, life moves into it. And advanced civilizations do not exactly needs rare habitats. A star and a bunch of common elements is enough to build and power habitats, simulation computers or whatever. We know this because we do it, and advanced civilizations won't be worse than us at this. So the question can be put this way: Why would advanced life refuse to expand into hospitable and empty habitats when they are available? This has no precedent in our experience with life, and requires some extraordinary explanation which I cannot even begin to guess at.
Finally, a company makes a product that I've been crying out for since ages ago. Now there just needs to be a consumer version of this video+USB-over-fibler protocol. It should, however work this way: the graphics processing should be done at the machine, and the fiber optics cable should just carry the video signal. If a 20 meter cable and a fiber-to-DVI+USB junction cost, say, $150 (not unrealistic), it could kick off the next mass revolution in home computing, where the computer itself becomes an appliance like a water heater.
Here's what I'm pictuing: People spend tons of money to make their computer quiet and well-cooled. But if the thing lived in the cold basement, they could bolt in cheap gigantic fans and disturb noone. But here's the kicker: The basement computer would be a multi-user system, where all the users of the household (including, for example, the living room display) would be using the same system simultaneously. Their rooms would contain displays and input devices only, wired in by fiber. Even if that happens, they're unlikely to get in each others' way, since by then these things will have at least 16 processor cores for them to share. But it means that if a single user needs to do something processor-intensive, she'll have the power of a pretty serious 16-core machine behind her, while the kids browsing myspace from the same computer (but on a different display) won't even notice.
3D GPUs are also about to go seriously multicore, and resource division on those will be easier than it is with CPUs. So if there are two gamers in the house, they could share a powerful multicore card and get acceptable performance. But if only one of them is playing, he can hog the resources of all the cores, and turn everything up to eleven.
This paradigm of the basement computing appliance could revolutionize the way hardware is made and marketed. Multisocket motherboards for the mass market could easily become common, but I'm picturing also a system of arbitrary daughterboards with extra processing units, which will speed up the system without forcing the owner to scrap things. Sure it will become a giant lego-like mess that sounds like a jet, but that's OK. It's in the basement (and will by then hopefully have sane power management which will turn off absolutely every part of every chip which isn't being used).
My point is that normal households with multiple computers today duplicate a lot of resources which go wasted, since single user has the opportunity to use them all simultaneously. The way to fix that is to pool all the household's processing into a single, big, arbitrarily extensible machine which stays out of people's way. And for that, we need a good long-run digital video over fiber standard. And maybe, with all the excess heat these things will put off, they really could double as the hot water heater!
Now Postini will get access to the Googlebrain!
on
Google to Acquire Postini
·
· Score: 2, Interesting
First of all, as many have noted, the point of this is not to adapt Postini tools to Gmail. That may happen eventually, but it's not the point. The point is that Postini offers enterprise services that Google never did, and already has a prominent userbase. And if you ask why enterprises don't just switch to Gmail and get the same spam-filtering for free, you don't understand how enterprises work.
I don't doubt that some of the spam filtering procedure developed by Postini will eventually help filter Gmail. Indeed, it wouldn't make sense in the long run for Google to keep two separate spam filtering platforms. But here's the point: the primary beneficiary of the buyout will be the Postini spam filter itself, the thing that will be sold for subscription fees to enterprises. That product will improve for one simple reason: Access to the incredible amount of data that Google has access to. We all help Google when we're kind enough to press the "mark as spam" button in Gmail. And I'm sure they remember, and our entry sharpens up whatever Bayesian algorithm Google uses to detect future spam. When Google's data merges with Postini's data, it will be very hard for other enterprise spam filtering providers to offer a product of similar effectiveness. To do so, they would need to store their own databases on a scale large enough to compete with Google - which isn't cheap. It is cheap for Google to supply Postini filters with raw data, since Google collect that data anyway. So Postini the pay service gets an incredible competitive advantage though it's intergration with the Googlebrain. That's not to mention the extra mindshare that the Google brand brings with it.
For those of us who wondered how Google plans to profit from all this investment in a free email service, this is a part of the answer: There will be a for-pay enterprise version based on the same investment. The same goes for Search, btw. So pay attention: this is Google trying to become something more than an ad pusher. And it's not a dumb idea: the marginal cost for Google to develop a good for-pay spam filtering system is small compared to the money they could sell it for.
And since you can already buy Google computers to search your enterprise for internal data, and those Google computers are heavily based on work Google developed for other goals (and for free access), we might ask the following question: What other things is Google good at, and would enterprises be interested in paying for products based on those skills? Google maps? For sure! But consider Google News, the human-free, smart organizer of articles by subject, relevance and prominence. Are there companies with a lot of data that could benefit from the sort of organization alorithms that run Google News? Damn right! Each year more enterprises are finding that the cheapness of data storage left them with attics of archival data that's a complete mess. I think we're starting to understand the "???" that separated Google's free services and Profit.
I've been around Slashdot long enough to have seen "Linux poised to explode on the desktop this year" articles. They were dense when KDE hit 1.0 in the mid 90's, and later when Sun released the codebase for StarOffice. When Linux stocks were selling like crazy and Eric Raymond was writing rants about how he's filthy rich but don't ask him for money, it was taken as axiomatic that a big proportion of couputer users would be running Linux. The thing is, those previous years really did seem special, so similar articles then seemed more plausible.
What do we have this year that would drive people to Linux? Nothing fundamental that wasn't there a year ago. By now there are a ton of copies of Windows XP, which is actually a pretty good operating system once supplemented with third party gear. Vast majorities of desktop users don't see any reason to get a new operating system, much less a new kind of operating system. That hasn't changed either. So, let's face it, this will most assuredly not be the year of Linux on the desktop.
I know what will happen next. The Gates Foundation will suddenly realize that poor kids deserve better than a French Linux, and they will generously donate a distribution of Windows for these things. You know, lock them in when they're young and poor, or as they would say, get them up to date with the world's software standards.
The only reason I can thing of for why they haven't done this yet: They'd have to release a streamlined version of windows without all the usual bloat, and hackers would very quickly turn it into the Windows that the rest of us always wanted.
I know that game is ancient and I'd love to see it modernized, but even as it is, it really offers the most replay value of any game I've ever owned. With the map editor you can create challenges for yourself that require some serious strategic thinking, and the variety of gameplay situations is basically limitless. Designing a map doesn't take very long compared to playing it, which is an important advantage over a game like NWN where this ratio is turned on its head.
What's great about DKII is also the option to play it like a god sim game, where you design something with the objective of establishing a more perfect harmony rather than just winning. That's a very interesting architectural challenge given the population of stange folks that you need to cater to, and all the different ways they can be made unhappy.
I long for a Dungeon Keeper game that's updated for the 21st century, one where I can write my own, more complex AI for the creatures in the game. DKII's AI is really quite good and makes things fun, but that's one thing that could really be beefed up for today's stronger computers.
If there were an OSS project to remake Dungeon Keeper with one of the modern free 3D engines, I would absolutely love to participate. I write pretty good AI code. I would never finish my dissertation, but whatever, it would be worth it! And yes, I would do this on a desert island or in jail, if I had to.
I also have many ideas about what could be added to the game to make it fun. Some of the things: Creatures would respond to their surroundings by decorating their space, getting into fights with creatures they hate, make a mess, clean up the mess of others, cooperate while fighting, etc. With a few improvements, this could become the greatest game ever made. I'm serious!
Slashdot Mirror
So what ads would I get? That might actually be interesting: Probably ads about recently released physics books, about how Neurosis will be playing in Buffalo (they'd know where I am!), about a comic book convention in Rochester, a sale on my favorite strings at the Guitar Center... - and I'm sitting here wondering whether this would be so bad. After all, these ads would each very likely result in me spending money, greatly increasing the efficiency of the whole advertising system. Potentially, this would mean that I wouldn't need to watch as many ads. Once they collect enough data, they would realize that most of them are a complete waste anyway, and that it would be wiser for them to just not bother with them. Also, this would really lower the economic barrier for submitting an ad to the media. It might be profitable for a local bar to actually advertise their decision to have Wednesday night trivia, because only people who live in the neighborhood, go to bars and like trivia would get the ad.
Even better would be this: They let me select an "I'm not interested in this sort of thing" option along with ads they show. Of course they wouldn't be contractually obligated to respect all my input, but... generally that sort of input would be good for both the consumer and the advertiser.
Agreed, and this is great news, because that sort of GPU will be so CPU-like that it will definitely also be able to do all the other stuff (besides games) that today's computers still do far too slowly: Media encoding and image/video rendering and processing. In all other categories, a bargain computer of today is plenty fast enough so that speedups will be almost imperceptible. I fully expect future 1080p video encoding to be done on super-parallel multicore CPU's whose main role will be to render games. And I expect it to be done very very fast, much faster than realtime even with very complex codecs.
I hate GUIs too, and you can be sure I'll be lobbying for a command-line-only interface for the Gimp. It might have a steep learning curve, but can you imagine how powerful and efficient that would be?
This leaves the impression that they haven't even worked out what this manual charging method will be, and might be leaving it for future editions of the OLPC. If this is wrong, can someone post a link that shows the manual charging system?
Wouldn't it be pretty ironic if they ended up using cheap child labor to make these?
Thank you for that fine and insightful rant. I clicked on the comments page thinking I'd have to compose something similar, but instead I think I will go grab lunch.
(And seriously, won't there be a Windows version of the open source driver? And if so, might it not surpass AMD's own Windows driver? And might this not be a part of AMD's strategy, out of recognition that everyone disses their drivers and that their coders cost them too many salaries? Probably not, but this will certainly do a lot to make people finally reconsider the undying meme that ATi cards are better but their drivers suck.)
In order to help give substance to the debate, Roman coded together some proof-of-concept stuff, but instead of his architectural ideas being looked at seriously and critically, Ingo instructs him to strip away most things and "well use it." That really should seem to everyone on the sidelines like Roman's ideas are being ignored without debate. Now, maybe Ingo is polite, Roman's work just sucks, and Ingo won't confront him on it. But if that's not the case, maybe there should be a (non-flamey) debate about the best architecture for the scheduler.
This means that Nokia can spend far less on advertising, so offer their phone for less and on more reasonable terms. For example, they can let carriers subsidize the price as a reward for signing a contract, so to the customer, the iPhone will look twice as expensive as Nokia's potentially superior knockoff.
It's a brilliant move, and it costs Nokia almost nothing - minimal advertising and no wasted engineering effort, since every major phone company needs to develop a modern touch-screen interface anyway. Might as well catch a free ride on the Apple hype-wave while they're at it!
But do they ever get to exactly the same value? Suppose they start off at .2, .4 and .7 and they get amped up to .6, .7 and .9. You should be able to re-expand these to their original values, right? Does it really ever happen that signals which start at different values get squashed to exactly the same value?
I agree about how it should be done, but I'd be surprised if the original dynamic range was unrecoverable from the modern recording. It's just that the differences in the levels were shrunk. They weren't completely destroyed. That's why I think they could potentially be re-expanded, and my question was: Why is there no common hardware or software that does this?
I don't think that's right. I worked at a radio station and we had our own compressor that our signal got fed through. So did every single other FM station. There is no need at all for the CDs to have this compression on them, since the destruction of the dynamic range is very easy to do for a radio station.
So now my question is this: Why the hell are we not doing that now, with this music-squashing epidemic in full force? The old arguments for why we should squash the dynamic range no longer hold. One of them was to increase the average volume compared to the hiss. Well, we can control hiss better than ever. Another reason is that amplifiers that play music which is on average quieter need to be turned up louder, which adds noise and distortion. But amplifier circuits are much better now, and this effect is tiny.
It just can't be that hard, from a caclulation point of view, to re-expand the dynamic range in a squashed recording. Real-time sotwtware should be able to do this easily. Yet I don't know of any such plugin. Google revealed that a shareware plugin called DFX ($20) might do something like this, but the details are unclear. Am I missing something? What's holding back the dynamic range re-expanders that so many of us want?
I'm pretty skeptical though. If evolving structures are so common that we see them even in a low-powered simulation, and every single star has so much freaking plasma, where are our plasma overlords? Or maybe that's hell, and those structures are just ... the souls of the damned! Oooh!
I thought of the same article. We need an explanation for why genetic diversity decreased so much in Europe and not in Africa (for example). And the mechanism of the wealthy families replacing the poor is a very plausible suggestion, since it has been independently shown to occur in Europe and not elsewhere. Of course the plague had something to do with it too, but that's not enough to explain the whole effect. It's important that the same population pressure applied for more than 30 generations, and that's long enough for some pretty impressive consequences to emerge.
The whole point of the industrial revolution was to reduce the proportion of people needded to work the land. It's not just about harvesting. Ever tried hand-thrashing grain with a flail? Neither have I, but it looks pretty labor intensive. Much harder than firing up a steam-powered thrasher. Sure it takes work to make one, an entire village can use it. So if you want to provide an answer for why these things were not made earlier, the answer cannot be that there weren't enough people for it, as though industrializing takes extra work. Industrializing saves work. And the process is continuing... and will pick up as AI and expert systems get better.
It's not quite that "all the lazy and stupid people died out" - it's more that the families disposed to working long hours and delaying gratification were the ones that, in each of 30 generations, had a survival advantage. And 30 generations of the same selection pressure really could make a lasting difference. But what the argument needs is evidence that in places like China, the same population pressures did not apply. Otherwise, he's not offering an explanation of why the industrial revolution was spawned in Europe and Asia didn't have its own home-grown version, despite a favorable climate and plenty of coal.
Of course, Marx was different in this way: He did make one prediction about the future whose means of production were unknown to him: he thought there would be a people's revolution in which people would take control of the technology developed in the capitalist era, because of the inevitable resort to artificial scarcity that the capitalist system will increasingly have to turn to. Scarcity will need to be artificial because technology will be able to meet all the basic and many of the advanced needs of everyone in the world. Capitalism doesn't work in situations of plentitude, so there is no market for breathable air (yet). So the artificial scarcity that Capitalists will need to create will eventually get so ridiculous that people will just depose them. As far as futurism goes, I think this outline is aging rather well.
And by the way, this is much closer to what Marx actually said than what most "Communists" claim he said. The Marx I read never advocated a revolution, resource distibution, or any of that other socialist stuff. He was a dialectician who thought that history has an inner logic and moves forward inevitably. Pleading with people doesn't move history; technology moves history. He argued pretty forcefully that Capitalism isn't the final system, but not because he was trying to stir up a revolution. It was just to convince people that it can't last, that, like every earlier technological/ideological era, it will be undone by the tools it eventually creates. So if Capitalism creates automatic strawberry harvesters because Mexicans get too expensive, and intelligent robots and fusion powerplants and workerless factories, it will eventually make the gear of it's own demise. Marx repeatedly extolled Capitalism for being so damn good at producing new technology in the most efficient way possible. It was Lenin, not Marx, who thought that a society can leap past all the stages of industrial and post-industrial capitalism and start a revolution with just an ideological vanguard. Obviously, that didn't work out. Marx was clear that technology drives ideology and not the other way around.
As far as consciousness retreating into virtual worlds, I find this rather probable. However, if civilizations in our galaxy have done this, we would notice it. Surely, some of the simulations would require increasing resources to maintain, because simulations get better the larger and more detailed they are. Give it a few years of increasing energy demands (say a million) and you should see solar-system-scale engineering projects, and then projects that go beyond their original solar system.
I'm trying not to make too many assumptions about life in general, but one safe one is that when habitat is available, life moves into it. And advanced civilizations do not exactly needs rare habitats. A star and a bunch of common elements is enough to build and power habitats, simulation computers or whatever. We know this because we do it, and advanced civilizations won't be worse than us at this. So the question can be put this way: Why would advanced life refuse to expand into hospitable and empty habitats when they are available? This has no precedent in our experience with life, and requires some extraordinary explanation which I cannot even begin to guess at.
Good point! Thank you.
Here's what I'm pictuing: People spend tons of money to make their computer quiet and well-cooled. But if the thing lived in the cold basement, they could bolt in cheap gigantic fans and disturb noone. But here's the kicker: The basement computer would be a multi-user system, where all the users of the household (including, for example, the living room display) would be using the same system simultaneously. Their rooms would contain displays and input devices only, wired in by fiber. Even if that happens, they're unlikely to get in each others' way, since by then these things will have at least 16 processor cores for them to share. But it means that if a single user needs to do something processor-intensive, she'll have the power of a pretty serious 16-core machine behind her, while the kids browsing myspace from the same computer (but on a different display) won't even notice.
3D GPUs are also about to go seriously multicore, and resource division on those will be easier than it is with CPUs. So if there are two gamers in the house, they could share a powerful multicore card and get acceptable performance. But if only one of them is playing, he can hog the resources of all the cores, and turn everything up to eleven.
This paradigm of the basement computing appliance could revolutionize the way hardware is made and marketed. Multisocket motherboards for the mass market could easily become common, but I'm picturing also a system of arbitrary daughterboards with extra processing units, which will speed up the system without forcing the owner to scrap things. Sure it will become a giant lego-like mess that sounds like a jet, but that's OK. It's in the basement (and will by then hopefully have sane power management which will turn off absolutely every part of every chip which isn't being used).
My point is that normal households with multiple computers today duplicate a lot of resources which go wasted, since single user has the opportunity to use them all simultaneously. The way to fix that is to pool all the household's processing into a single, big, arbitrarily extensible machine which stays out of people's way. And for that, we need a good long-run digital video over fiber standard. And maybe, with all the excess heat these things will put off, they really could double as the hot water heater!
I don't doubt that some of the spam filtering procedure developed by Postini will eventually help filter Gmail. Indeed, it wouldn't make sense in the long run for Google to keep two separate spam filtering platforms. But here's the point: the primary beneficiary of the buyout will be the Postini spam filter itself, the thing that will be sold for subscription fees to enterprises. That product will improve for one simple reason: Access to the incredible amount of data that Google has access to. We all help Google when we're kind enough to press the "mark as spam" button in Gmail. And I'm sure they remember, and our entry sharpens up whatever Bayesian algorithm Google uses to detect future spam. When Google's data merges with Postini's data, it will be very hard for other enterprise spam filtering providers to offer a product of similar effectiveness. To do so, they would need to store their own databases on a scale large enough to compete with Google - which isn't cheap. It is cheap for Google to supply Postini filters with raw data, since Google collect that data anyway. So Postini the pay service gets an incredible competitive advantage though it's intergration with the Googlebrain. That's not to mention the extra mindshare that the Google brand brings with it.
For those of us who wondered how Google plans to profit from all this investment in a free email service, this is a part of the answer: There will be a for-pay enterprise version based on the same investment. The same goes for Search, btw. So pay attention: this is Google trying to become something more than an ad pusher. And it's not a dumb idea: the marginal cost for Google to develop a good for-pay spam filtering system is small compared to the money they could sell it for.
And since you can already buy Google computers to search your enterprise for internal data, and those Google computers are heavily based on work Google developed for other goals (and for free access), we might ask the following question: What other things is Google good at, and would enterprises be interested in paying for products based on those skills? Google maps? For sure! But consider Google News, the human-free, smart organizer of articles by subject, relevance and prominence. Are there companies with a lot of data that could benefit from the sort of organization alorithms that run Google News? Damn right! Each year more enterprises are finding that the cheapness of data storage left them with attics of archival data that's a complete mess. I think we're starting to understand the "???" that separated Google's free services and Profit.
What do we have this year that would drive people to Linux? Nothing fundamental that wasn't there a year ago. By now there are a ton of copies of Windows XP, which is actually a pretty good operating system once supplemented with third party gear. Vast majorities of desktop users don't see any reason to get a new operating system, much less a new kind of operating system. That hasn't changed either. So, let's face it, this will most assuredly not be the year of Linux on the desktop.
The only reason I can thing of for why they haven't done this yet: They'd have to release a streamlined version of windows without all the usual bloat, and hackers would very quickly turn it into the Windows that the rest of us always wanted.
What's great about DKII is also the option to play it like a god sim game, where you design something with the objective of establishing a more perfect harmony rather than just winning. That's a very interesting architectural challenge given the population of stange folks that you need to cater to, and all the different ways they can be made unhappy.
I long for a Dungeon Keeper game that's updated for the 21st century, one where I can write my own, more complex AI for the creatures in the game. DKII's AI is really quite good and makes things fun, but that's one thing that could really be beefed up for today's stronger computers.
If there were an OSS project to remake Dungeon Keeper with one of the modern free 3D engines, I would absolutely love to participate. I write pretty good AI code. I would never finish my dissertation, but whatever, it would be worth it! And yes, I would do this on a desert island or in jail, if I had to.
I also have many ideas about what could be added to the game to make it fun. Some of the things: Creatures would respond to their surroundings by decorating their space, getting into fights with creatures they hate, make a mess, clean up the mess of others, cooperate while fighting, etc. With a few improvements, this could become the greatest game ever made. I'm serious!