"Tracking the RNG" would help you win the game, but it doesn't tell you anything about how to play the game.
That would be my point.
This AI learns to play the game, it then wins the game using experience it gains in the same way a human does - feedback from the game score.
That is one possible interpretation, which is not supported by the statements so far. That is not to say that it is not the case, only that it is not currently supported by what I have seen so far; something along the lines of "We tested this against games with multiple RNGs with no perceptible change in AI performance" would support that interpretation. There are other interpretations. People are *assuming* that "wins" = "plays the game" - and the company that did it isn't relieving anybody of that perception (understandably). That's the point. Exploration of other explanations for success are warranted.
Consider that, for games which possess a weak RNG (i.e. predictable starting conditions and knowable changes in game play, i.e. most old console games), it is in theory possible to play *blind* - in other words, not actually paying attention to what's going on on the screen, but simply hitting buttons at precise enough intervals. If 'score' is taken as a proxy for 'how far you can get in the game' (ceteris paribus, someone with a higher score made it longer), then most known machine-learning methods will converge on that/those sequence(s) without any understanding of 'the game' per se. It may even be possible to do that for short gameplay sequences based on pattern matches to known game conditions. While that does get off into the semantic weeds of what 'playing the game' is, it is difficult to differentiate between an AI which has 'learned' to play the game in the sense that it understands abstract rules, interprets game state, and makes decisions about what to do based on that observed state, and a neural network which has converged on the correct list of keystrokes to pwn the computer given certain observed starting conditions. One of them is impressive; the other one isn't, quite so much.
I find myself wondering about the following question:
How did they differentiate "learning to play the game" from "learning how to track the game's RNG"?
Most video games have ridiculously simplistic PRNG generators embedded in them. An AI might get "sidetracked" and learn how to play the underlying RNG output of the game, rather than the game itself. That would yield really good results for most arcade games of this type, I imagine (weak RNG, limited input and timing options, etc.) I don't know if they checked for that possibility.
Easy way to check, though: Reach into the game and substitute a better RNG (cryptographically-strong/hardware/quantum) RNG for the one in the game. That would enable you to quickly determine the difference. If the AI's game performance suddenly goes to shit, it wasn't a real game-playing AI. If it doesn't, well, all hail Skynet, I guess.
"... We wrapped a robot in a dead sparrow and decided to see if we could fool the other sparrows into interacting with our creepy, ghoulish automaton! It's *science*!"
And of course, it was COMPLETELY UNEXPECTED that the grisly abomination stapled to a tree branch triggered aggressive reactions from the other sparrows. Because every living thing JUST LOVES to be confronted with a soulless golem wrapped in the dead flesh of another of its kind. And that never causes pants-shitting terror or anything.
I can see it now:
Sparrow 1: "OH MY GOD! IS THAT... *THING*... WEARING FRANK'S FACE? IS IT?! FRANK??!?!" Sparrow 2: "It's not him anymore. IT'S!...NOT!...HIM! IT'S A MACHINE! Help me destroy it! Be his egg-layer one last time!" Sparrow 1: "*snf* OK... OK... oh God, Frank... God help me..."
Yup. Science.
Is there, like, a review board or anything? Maybe that could screen some horror flicks before writing checks for this kind of bullshit? "New rule: If your study is substantially similar to the plot of any one of this library of 100 horror movies, or if it has a plausible chance of producing similar outcomes, we're not going to fund it."
"This would be most dramatic if the intellectual property was produced in one nation under its laws then used without license by another nation to effectively eliminate the benefits of the intellectual property protects."
Yeah, uh, you kind of miss the point here. Rothbard, "Fractional-Reserve Banking" and "Anatomy of the Bank Run", game over, you lose.
The problem with FRB: it constitutes fraud. While I maintain that this statement is prima-facie obvious, I have the feeling that it will escape some portion of the crowd, so we'll try a thought-experiment.
"Fractional reserve" banking says that for every, say, $1000 of demand deposits (ex. checking accounts), the bank need only keep some fraction available at any time, on the theory that not all $1000 will be demanded at once. Yes, that's sort-of true - statistically, most of the time, most people will be content to leave it in the bank, and only call it out as-needed. Most of these calls for money will, in turn, be deposited in another bank, thereby adding an extra level of "protection": inter-bank transfers can be "batched" and resolved on different time scales than the demand-deposit processing (e.g. accounts squared at end-of-day, end-of-week, whatever).
However, what the bank is saying when you deposit money in a demand-deposit account - "your money is available for withdrawal at any time of your choosing" - is, literally, not true. *Your* money has disappeared into someone else's pocket, in that it has been loaned out to some other party as soon as it hits the bank. This is not a "theoretical" untruth - it is a real untruth, in that at all times, the bank is illiquid, i.e. does not possess sufficient capital to redeem all its demand-deposit accounts on actual demand. This is what is referred to in any other instance as "constructive fraud". It is useful to compare this to the eponymous Ponzi scheme, where the fraud consists of there being no actual capital or investment to satisfy the promised payout schedule to current investors, requiring that new investors be found to service existing obligations.
This is not the worst feature of fractional-reserve banking, though. FRB is the gateway for the money multiplier and hence inflation; a bank with a reserve requirement of 5% (larger than the current reserve requirements, note; I believe they're hovering at less than 1%) can, with a deposit of $1000, immediately turn around and loan out $950 of that money. This functionally doubles the amount of money in the economy ($1000 of "fantasy" money, and $950 of "actual" money floating around). That $950 typically gets deposited in *another* fractional-reserve bank, almost invariably with the same reserve ratio (set by the central bank, and reinforced by consumer preferences; ceteris paribus, a higher reserve ratio implies lower interest rates on deposits, providing customers incentive to move their deposits to another bank), at which point the cycle starts again ($950 in "fantasy" money, of which $47.50 is kept on-hand, and the remaining $902.50 lent out). This multiplies that original $1000 of "real" money to something like $20,000 in terms of its real economic effect. SUPRISE INFLATIONSECKS LOL. Then, when you start printing up more money (cue the Fed) and tweaking reserve ratios (cue the Fed again), you wind up with - wait for it - more inflation. Inflation has well-known and universally-observed destructive effects, penalizing saving and encouraging increasing amounts of debt, since debt is paid off with future money that is worth less than the original loan. How's that working out for us so far?
(Side note: it is instructive to note the identity between fiat currency and counterfeit, with the only difference being the identity of the printer of money.)
Austrian economic theory consists of the "duh no-shit" observation that this has an effect on the economy - inflation makes money cheaper to obtain (hey, they print it for nothing!), thereby making marginal enterprises "profitable" under inflationary conditions. This produces the "boom". When those conditions cease - there is an lower bound on the worthlessness of fiat currency, beyond which it is not used except as kindling - all of those bad ideas *come home to roost*, with the acc
Or, what potential patent holders will have to do in twenty years is find their Sonny Bono and get the term of patent rights extended to the life of the Universe plus 99 years.
I wish that people would fucking figure out how this shit works already. Any process that makes use of government to shake somebody down ratchets in the direction of "more shakedown/less freedom" only, until enough people get sick of this state of affairs and begin collecting heads. This is independent of how good an idea you personally think it would be to shake some given person A down to benefit person B. Solution: refrain from shaking people down, don't start in the first place. (And yes, patent and copyright *are* very much shakedowns.)
Hayek. Mises. Rothbard. Read 'em. Learn 'em. Live 'em.
Neither God nor the Universe are under any obligation to "make sense" at all, much less to you, much less any kind of sense that you would identify as such.
I'm frankly surprised that nobody has mentioned the #1 noise source and probable steganographic message carrier out there: spam. It's ubiquitous, customarily comes with a shitload of SEEMINGLY random strings whose purpose is ostensibly to confuse hash-based and keyword filtering (but which could contain God-knows-what), is easy to do, and doesn't raise any eyebrows. What do most people do with spam? Throw it in the trashcan, of course, they can't hardly get rid of it fast enough. You can scatter it across millions of email address, camouflaging the one you're really sending it to. And only for those with the secret decoder ring would the funny strings have any meaning...
You are making the assumption that "winning" in a market is an absorptive process. Yes, the sequence looks like this:
1) Use larger store of cash to cut all competitors out of the market. 2) Once all other competitors are out of the market, raise prices ("gouge"). 3) Profit!
Great Slashdot economic analysis, but you forgot steps 4 and 5:
4a) Watch as old competitors re-enter the market or new competitors arise due to the _demonstrated profit potential of the market_ (remember, I'm gouging, so I'm making Big Bux)
or
4b) Watch as customers, disgusted at high and rising prices, decide they don't want what I'm selling and go find something better to do with their money.
and
5) Go out of business.
Rinse, repeat. The only way to prevent steps 4) and 5) is to ensure that your competitors never arise (using regulations and special privilege backed by deadly force to lock out competitors, aka Microsoft, Ma Bell, USPS), and/or to ensure that customers cannot stop buying your product if they are unhappy with it (i.e. government of all stripes).
Nothing is ever entirely static. People who raise prices in a market make that market more attractive for competitors, not less.
Regulations only serve to prevent the entry of competitors into the market (otherwise, what is the point of a regulation?), allowing existing players to raise their prices - or keep them at a high level - without fear of competition. Sometimes those price raises only cover the actual raise in costs due to the regulation. Mostly they're a lot bigger, since the businessman is guaranteed a lower incidence of competition in his market.
Murray Rothbard. Ludwig von Mises. F.A. Hayek. And many others.
Basic truth: Government cannot interfere in a free market - IN ANY WAY - without distorting it. What is the free market? A free market is a market where buyers and sellers are able to meet and make a trade without interference. This trade is mutually beneficial, otherwise it would not have been made. When government interferes in this arrangement, these trades are either not made, or they not as beneficial to both parties as they otherwise would have been. (Otherwise, what would have been the point of the interference? The "best thing" would have happened anyway, making government intervention a total waste.)
You are correct in that government benefits _some_ business. Microsoft, for example, made billions of dollars based on the notional value of its intellectual property, which government secured _at no cost to Microsoft_. You are statutorily unable to copy the software, disassemble or reverse-engineer it, or remove pieces from it to use in your software, without being guilty of a crime. I find it droll that a company which took maximal advantage of this amazing boon now expresses indignance at being charged with "predatory pricing", as though such a thing were possible. (Lie down with dogs, wake up with fleas, etc.) You can examine the effects of government on the market in other industries as well, creating what are commonly known as "cartels".
Enforcement of contracts presupposes a situation where people make and break promises willy-nilly, which _of course_ requires the Hand of God^H^H^HGovernment to correct, forcing everybody to "play by the rules". There is no such epidemic of perfidy about; those people who do not honor their contracts typically discover many fewer takers for their next proposed contract. The problem is self-correcting. Further, even in those instances where a contract must be "enforced", it is a highly dubious proposition that government must be involved; examine the practice of private debt collection as one example, a case where one party made a contract with another and then refused to make good on their obligation. There are a multitude of ways to seek restitution; government intervention is not the only path, and indeed is hardly the best.
There is no such thing as a "pure public good". "Pure public goods" are a misnomer; they inevitably wind up "belonging" to a certain elite class at the continuing expense of the rest of the populace. Something "belongs" to you only if you have complete control over its disposition; if you believe that you "own" any portion of purportedly public property, try taking your aliquot piece of it and using it for your own purposes and see how far that gets you.
Finally, governments do not "bust up" monopolies, they tend to create them. Reread above; Microsoft was created by the government enforcement of copyright of software. If you doubt this analysis, ask yourself this; what are the effects of regulation on an industry? They constitute a statutory (i.e. "non-market-based") barrier to entry to that market thereby creating an unnatural shortage of competition, of course. Monopoly - total control of a particular industry - is a natural extension of regulation. Indeed, once traced back to the root, all known instances of monopoly have sprung forth from government intervention in a market.
You need a lesson or two in economics, I'm afraid. You could also stand a lesson in the precise definition of "anarchy". I daresay you'd most benefit from a heapin' helpin' of Shut The Fuck Up, too, but I have my doubts that you'll partake. Oh well. Have a nice day!
The governments of these African countries, like the government of India before them, are in the process of subsidizing the development of what is perceived to be a cash cow of limitless milkability, IT. This process is nothing more and nothing less than seizing money at gunpoint from other, more productive domestic industry (natural resources development as one example) or getting it from dumber countries (like, say, the US and its billions of dollars of foreign aid, ironically likewise looted from the American taxpayer), and giving it to another industry to make it grow in defiance of market forces. Governments are subsidizing the production of millions of PhDs, handing out favors to "tech-savvy" "entrepreneurs" and foreign companies to take advantage of the perceived riches of the tech industry, not realizing a couple of very basic tenets of economics:
ALL OTHER THINGS BEING EQUAL, WHEN SUPPLY GOES UP, PROFITS (AND PRICES) GO DOWN.
and
IT IS NOT POSSIBLE TO CIRCUMVENT MARKET FORCES. USING GOVERNMENT TO FORCE THE ISSUE LOOKS BETTER NOW, BUT COSTS MORE LATER.
The problem is, this is not an endless phenomenon. It wasn't profitable to locate things in India before, for a multitude of reasons (lack of infrastructure, lack of education, social problems, whatever). It will likewise be unprofitable in the future, when their millions of PhDs are hacking cabs in New Delhi to make rent, or becoming farmers. (You can see this process beginning now. The market there has reached capacity, and other places - like Africa, Land of Ceaseless Warfare, Spam, and Disease - are being seriously considered as places to invest in tech, because the market in India is getting too inflated.) It sure as hell has been unprofitable and/or just plain dumb to locate any form of tech industry capital in basically any African country, where the odds of its being nationalized, destroyed, or devalued in the customary and predictable political upheaval are astronomical.
The cornucopia of benefit from IT and tech in general is mostly illusory. It came about in the US largely through a government/Federal Reserve easy-credit policy in the 90s that allowed all manner of idiocy to get funding and look great on paper (AKA the dot-com boom - pets.com, anyone?), followed by the bust when all of these crappy investments based on bullshit were exposed as the stupid ideas that they were. Yes, there is some benefit to tech, as long as it enhances productivity and quality of life. No, its benefit on life and productivity are not infinite, nor is this benefit anywhere near as bountiful as some think. It seems that the governments of other countries, enthralled by the idea of a trillion-dollar business tax base (or "loot pond") springing up overnight with a minimum of effort, are going to go down this same road with precisely the same heartbreak at its end. The citizens of these countries would do better to leave their neighbors alone and spend their time farming and defending their property from invaders. After a few decades of respect of property rights and natural rights have set in, then they could begin working their way up the industrial/informational ladder, and would be in a much better positioin than we are now. (For that matter, we in the US should probably take the same advice.)
You're missing the point, which is entirely usual for an AC. The response in the FAQ/QnA to all questions of "How will we handle X?", where X is "unbonded/poor/anonymous/whatever senders", "mailing lists", etc. have the same two answers: "Well, you'll have to put them on your whitelist", or "you'll have to look at the messages individually to determine if you want to read them".
BUT WE DO THAT NOW. There is nothing new here, except an elaborately-designed bit of wankery designed to flag potential spam messages in a slightly novel way. You still have to undertake the Aegean task of discovering which ones are spam and which are not, unless you are willing to forego communication with everybody who is not on your whitelist and/or who does not post a sufficiently large bond. The very best you can hope for from this system is to recover some amount of money for reading spam, THEORETICALLY. (The ability of this service to guarantee this has not been demonstrated to my satisfaction, any more than the existing system's capability to prevent spammers from obtaining access in the first place.)
To address some points of your "rebuttal":
"(x) Mailing lists and other legitimate email uses would be affected Incorrect, see other posts"
No, not incorrect. It increases the difficulty of running a mailing list, to say nothing of increasing its costs.
"(x) Users of email will not put up with it Depends on the effectiveness and the cost. This system promises legitimate users negative cost!"
Hell, I can go through my spam archive and find messages promising me wealth, a 50-foot penis, and hot sex with teen virgins. I don't believe that, either. (Wait until that legitimate user gets every bond redeemed due to losing the eternal online popularity contest, or even better, has his bond account emptied due to a security flaw somewhere. We'll see what he/she thinks of it then.)
"(x) Requires immediate total cooperation from everybody at once Incorrect, a non-participating recipient will simply not request bonds. As with most other anti-spam solution, a fall back address can be used which is checked with lower priority and stricter content rules to discourage users from sending mail to the non-participating address."
Thereby providing, for 99% of all email users, zero benefit over the existing system. (A "solution" where you still have to clean a spam trap is not a solution.)
"(x) Armies of worm riddled broadband-connected Windows boxes Correct, a working and secure micropayment system is a requirement."
... Which we don't have now, and for various reasons, will probably _never_ have. Why not predicate the existence of a functional antispam system on something vaguely probable, like a secret group of four superhumans wearing leotards? We could call them the "Spamtastic 4", and they could fly around preventing evil spammers from spamming. Ooo, and they could have a moon base too. Yeah, that's the ticket.
"(x) Joe jobs and/or identity theft Incorrect. Why would you accept to provide bonds for messages which you didn't send?"
I wasn't aware that you had to agree to an identity thief/cracker emptying your bond account of however much you had in it. A security flaw hardly requires your permission to exploit. (But now that I know you have to agree to it, that makes it much better. Tell me, do the attacks against the system require the use of the RFC 3514 "Evil Bit"?) For extra crunchy added badness, ponder possibilities like bond accounts being attached to checking accounts, and the average state of home network security. ("Hey, I'm gonna go wardriving and spam myself a new Mercedes!")
"(x) Bandwidth costs that are unaffected by client filtering Incorrect. With increasing popularity of Atte
(As a side note, what happens if you receive mail without an associated bond? 12.2Q in the Q&A says "Well, you could still read it", which OBVIATES THE ENTIRE FUCKING POINT!!! Yet another idiotic spam "solution", in other words. Oh well. Here's where it scores on the Spam Solution Checklist:)
approach to fighting spam. Your idea will not work. Here is why it won't work. (One or more of the following may apply to your particular idea, and it may have other flaws which used to vary from state to state before a bad federal law was passed.)
( ) Spammers can easily use it to harvest email addresses (x) Mailing lists and other legitimate email uses would be affected ( ) No one will be able to find the guy or collect the money ( ) It is defenseless against brute force attacks ( ) It will stop spam for two weeks and then we'll be stuck with it (x) Users of email will not put up with it (x) Microsoft will not put up with it ( ) The police will not put up with it (x) Requires too much cooperation from spammers (x) Requires immediate total cooperation from everybody at once (x) Many email users cannot afford to lose business or alienate potential employers ( ) Spammers don't care about invalid addresses in their lists ( ) Anyone could anonymously destroy anyone else's career or business
Specifically, your plan fails to account for
( ) Laws expressly prohibiting it ( ) Lack of centrally controlling authority for email ( ) Open relays in foreign countries ( ) Ease of searching tiny alphanumeric address space of all email addresses (x) Asshats ( ) Jurisdictional problems (x) Unpopularity of weird new taxes ( ) Public reluctance to accept weird new forms of money ( ) Huge existing software investment in SMTP ( ) Susceptibility of protocols other than SMTP to attack ( ) Willingness of users to install OS patches received by email (x) Armies of worm riddled broadband-connected Windows boxes ( ) Eternal arms race involved in all filtering approaches (x) Extreme profitability of spam (x) Joe jobs and/or identity theft ( ) Technically illiterate politicians (x) Extreme stupidity on the part of people who do business with spammers ( ) Dishonesty on the part of spammers themselves (x) Bandwidth costs that are unaffected by client filtering ( ) Outlook
and the following philosophical objections may also apply:
(x) Ideas similar to yours are easy to come up with, yet none have ever been shown practical ( ) Any scheme based on opt-out is unacceptable ( ) SMTP headers should not be the subject of legislation (x) Blacklists suck (x) Whitelists suck ( ) We should be able to talk about Viagra without being censored ( ) Countermeasures should not involve wire fraud or credit card fraud ( ) Countermeasures should not involve sabotage of public networks (x) Countermeasures must work if phased in gradually (x) Sending email should be free ( ) Why should we have to trust you and your servers? ( ) Incompatiblity with open source or open source licenses ( ) Feel-good measures do nothing to solve the problem ( ) Temporary/one-time email addresses are cumbersome ( ) I don't want the government reading my email ( ) Killing them that way is not slow and painful enough
Furthermore, this is what I think about you:
( ) Sorry dude, but I don't think it would work. (x) This is a stupid idea, and you're a stupid person for suggesting it. ( ) Nice try, assh0le! I'm going to find out where you live and burn your house down!
Great. So AMD has apparently rediscovered the Harvard architecture, or a variant thereof. (The "virtual" Harvard architecture?) [ http://en.wikipedia.org/wiki/Harvard_architecture ]
Has it been adequately demonstrated yet that there is in fact nothing new under the sun? No? Okay.
For a bonus laugh, I wonder if they're going to get a patent on this "novel technology".
Hmm. Three stories down, it appears as though Intel has a demonstration of "superfast communications technology in excess of 2 Gb/s" next week.
Coincidence? I think not.
Sounds like someone at Intel forgot to do their mental Kegel exercises. "It's done! It's amazing! It lives!" "Umm, wait a minute. How can we actually _demonstrate_ this?"
I suppose turning the convention hall into a 32,000x32,000 full-surround-sound 60-fps digital pornograph would be singularly unwelcome at the developer's conference, but... ah, wait. Never mind. (Only problem is, do they have enough time to do it? And do I have enough time to attend? Problems, problems...)
First of all, it would be "libel" if it is published in a written form. "Slander" entails _verbal_ communication. Second, truth is an affirmative defense for a charge of libel. What does that mean? It means that if what was said in the article was _provably true_, then it doesn't matter if there was any hostile or malicious intent on the part of the author. You cannot be fried for reporting fact. If "TheDeacon" has a foot fetish, and Joe Reporter reports that he has a foot fetish, whether or not he portrays that in a favorable or unfavorable light, it is _provably true_ and therefore is not libel. Third, being a private citizen is not a silver bullet to prevent libel. Yes, it is true that being a public citizen makes it nearly impossible to successfully sue for libel. That does _not_ mean that being a private citizen makes it _easy_ to successfully sue for libel. Someone else posted the three criteria for defining "libel" above, which must _all be met_, and which I encourage the reader to go back and review. I will concede that IANAL, but unless my memory core is getting seriously rusty these days, I know these preceding three points to be true nonetheless.
Lose the FUDstorm and pick up a a copy of the Reporter's Stylebook, Slashdot. And find a fucking dictionary while you're at it. It contains words and their _proper_ meanings.
What I _said_ was that this is not the new Big Iron, because a cluster of PCs does not qualify. It is not "engineered" in the sense that a Cray is, and I'm willing to bet that if you ask an actual Cray engineer you'll get your ass kicked for implying that some bozo on the street with a half a brain and a few bucks can build a supercomputer that will beat out something designed by a highly-trained engineering team from one of if not the best supercomputing manufacturers and research corporations.
To address the next point, nothing about the Flat Neighborhood Network makes it suitable for finely-grained problems, or at least as suitable as a "real" supercomputer. Throwing more switches at the problem and being clever with the wiring helps, but not to the extent that it will beat out a well-engineered chunk of Big Iron. Perhaps most importantly, there is a limit on the design of this network imposed by mathematics; discovering the proper "wiring" of the network is a nasty combinatorial problem, and is (according to them) difficult to solve for even 64 processors. Try doing it for 128 and observe the crispy crunchiness of this problem as you gouge out your own eyeballs and go insane. This type of cluster is not fundamentally different from a Beowulf cluster; while the result may well be useful for a certain class of problems, it's not worth getting all gooey about, primarily because there's _nothing new going on here_. Christ, if you want a big supercomputer, look at distributed.net. I bet they'll be able to beat 10 TFLOPS, at least theoretically, and d.n is yesterday's news. It just flat-out doesn't lend itself well to some problems of great interest. Yeah, it'll crack batches of keys just as slick as you please. What it won't do is something like weather modeling, at least not without bringing the Internet to a screeching halt as all available bandwidth is saturated with data requests.
Looking at the Cray site, they certainly are offering a cluster supercomputer. They're also offering several other types of supercomputer as well. As I said before - a Beowulf cluster is not the solution to all problems. Show us the benchmark on this bad boy for something nice and nasty, like turbulence modeling. I bet that all those processors choke nice and hard on the bottleneck and drop that 10 TFLOPS figure to something much more realistic, like about 1 TFLOP, in this circumstance.
The title says it all. Big Iron is _engineered_. No matter how big or how spiffy a Beowulf cluster is, it's still just a bunch of PC motherboards kludged together with a bunch of network cards. There is a reason Crays are expensive - they are _worth it_ from a performance standpoint, because not every problem lends itself easily to the solution of a Beowulf cluster. Some problems require the exchange of a lot of data between a lot of nodes, and a little math will show that it won't take much data interchange to saturate even a GigE switch. Adding more machines is not going to help; craftily designing and overengineering the network _might_, but by the time you get this whole damned thing glued together well enough to approximate a Cray's performance, you'll have spent enough to have just flat-out bought a Cray in the first place.
As others have noted, while this thing may have a theoretical peak performance of 10 TFLOPS, I'm willing to bet that number goes down like Monica Lewinsky on Quaaludes when you feed this magical supercomputer a problem that's _not_ suitable for distributed.net (i.e. one where computations on one node are dependent on computations on another node, like fluid-dynamics problems, turbulence, etc.)
Yeah, it's interesting as a curiosity, but this is by no means spectacular. Beowulf is good for what it's good for, which is a "poor-man's supercomputer" that works well for coarsely-parallel problems that don't require a lot of internode communication. It's not the Philosopher's Stone, folks.
The problem is that current advances in computing technology are focussing on "infrastructure"-level improvements; improvements in the actual hardware. A couple of companies (Apple) have spent money on the user interface. Noone, to my knowledge, has made a credible or successful effort to construct a metadata layer above commonly used information, indexing and sorting it so it is presented in the most effective fashion and linked to things that make sense. Notes:
a) I am _not_ talking about a language to do this in, because I'm sure some nitwit will chime in with "What about (XML|whatever TLA is the newest buzzword)?" That is "infrastructure". I am talking about actually _doing_ it, which hasn't been done in any global context yet.
b) Think Star Trek. Ever notice the funky engineering consoles that changed menus and key positions based on whatever information was being accessed? Playing with the warp-core antimatter injectors, or whatever, popped up a different screen and a different means of playing with it than firing the phasers. This is analogous to the process necessary to render information "usable"; Mr. Gelernter, although obviously floating off in a nether zone all his own, nevertheless gets this one right. People have to sit down and engineer default "pathways" through the morass of information out there with lavish effort and much cursing, not to mention a healthy dose of voodoo. The infrastructure to do this has to be ubiquitous, reliable, and standardized (more or less). Bill Gates' quote seems to be taken out of context; I think he's attempting to say that there are no intelligently designed links between most pieces of data that are stored on the average PC, and this is something that needs to change. As it so happens, he's right on that score.
This is not a new idea, and the author of this "whitepaper" is a twit.
Some time back on a mailing list far, far away, this got knocked around by myself and a couple of other people. (This was around the time of CR2.)
As I conjectured at the time, and still believe, the primary reason most worms fail is that they are either
a) written in a language with which 99% of programmers have very little skill (C, C++), or b) written in a remedial computer "language" that is useless to begin with (VB, macros, etc.)
This is true with every known worm to date. The problem with VB and languages of that type, well, never mind. If you can't figure it out yourself, stop reading now. The problem with C and C++ is that it is extremely easy to shoot yourself in the dick with either of these languages. (See the Morris worm's replication-rate bug for an example.) The problem is that with such a language, if you do not do everything exactly right, there will be a flaw in your worm and it will most likely fail, subtly if not spectacularly.
I conjectured that it would be a Good Thing (for he who wants to 0wn a large portion of the Internet) to investigate other languages instead. Prolog came immediately to mind. It seems ideally suited for this task as it is an intrinsically goal-oriented language. In addition, all the "complexity" of a Prolog program is hidden inside the resolution engine (which can be made quite small), so given that the resolution engine is operating correctly, if your program is well-designed, you have very few surprises to worry about. (Typically, although this is by far not a globally true statement, Prolog programs with bugs in them simply don't work, rather than failing subtly while appearing to work. That seems to make them a helluva lot easier to check.)
The fact that Prolog is an interpreted language makes it nice and portable; it is possible to create system-dependent and system-independent goals, and let the program figure out where it is and how to deal with it at runtime.
Also discussed for this bit of nastiness (I call it an "amoeba", after the slime mold which has similar traits) were the following:
a) goals added to the database from non-local sources would require a signature for each goal or batch thereof to prevent poison goals from being automatically propagated through the network (said signature using a private key, held by the Evil Overlord, matching a public key distributed with the worm);
b) "local" can be defined as being inside a clique of some finite size, say 10 computers, where goals and derived goals are propagated freely between members of the same clique, but not from members of different cliques unless they bear a valid signature;
c) infection can be maintained within a clique and progress to new machines from the "axolotl" effect; that is, machines will attempt to replace hosts in the clique that are no longer communicating, and at some random point, will decide all other hosts in the clique are "dead" and will seek to regrow the entire thing;
d) communication is probabilistic, not deterministic, since this is an effective means of frustrating attempts to locate infected machines; communications attempts are ignored at random, regardless of source or content, and spurious communications will be sent (random recipients, forged IPs, etc.)
e) Randomizing of opcodes using several different tricks, like instruction reordering a la Intel, instruction rewriting (essentially, replacing instructions with a set of different instructions that performs the same task), etc. The rewriting bit is essentially lambda calculus. >;-> Statistics for rewriting seem to be fairly nasty; I haven't had any luck coming up with an analytic expression for them, and have foregone a mathematical approach for a good old-fashioned "high-water/low-water" method for dealing with the fact that unless all isomorphic instruction strings are the same size (which they're not), this thing will have some serious problems with unbounded growth at some point.
These are a few of the ideas involved in this whole thing. Note carefully that none of this stuff is revolutionary; rather, it simply brings some old concepts back and puts them in a newer context. Once someone jerks their head out of their arse and figures out that C/C++/Java/whatever are not the best languages to do abso-fucking-lutely-everything in, this sort of thing will pop up and will be nastily difficult to defeat. (If indeed it's even discovered.)
I dunno. Waiting for a lengthy fsck can be very satisfying at times. Other times, I just want to get a quick fsck and go on about my business. (Typically around noon.) Depends on the day, I guess.
The advantages to such a system? Well, like you said, if you can afford to wait for a lengthy fsck once in a while, you're okay. If you just can't contain yourself, though, it might be worthwhile to go ahead and get it over with. The only problem with this is that if you absolutely cannot make it all the way through a slow fsck, you might pick up a reputation and/or an unflattering nickname ("Minuteman", for example.) It all depends on what you want; if instant gratification is important to you, you should be fine, modulo the occasional heckling and snickering by coworkers who are plugged into the office rumor mill. (But see touch(3); it may be more worth your while if you're all about speed as opposed to endurance, and might help with the problem of other people's cruel tendencies.)
As though 802.11 wasn't bad enough. Now we can have someone sniffing hard drive accesses as well?
I wonder when "Bluesnort" will be coming out. >;->
Does anyone know if the encryption for Bluetooth is as braindamaged as some of the others out there at the moment, or if it's actually something halfway decent?
the Chief Rabbinate of Israel, wanting to jump on the online-learning bandwagon, has designed a new curriculum. The University of Tel Aviv now offers a 90-day online course in circumcision, titled "Who wants to be a Mohel?"
Reception from Jewish communities has been lukewarm at best, especially as a typo in the online documentation for the course recently led to some accidents with unskilled graduates mistakenly "collecting tips", unaware that this refers to the custom of providing a gratuity for the mohel's services. Embarrassed by this, the leadership of the University of Tel Aviv has temporarily suspended the curriculum "for review".
This follows a rash of online-learning initiatives from respected universities, including Oxford's "eParasitology", Cornell University's "RealProctology", and Johns Hopkins "myGynecologist.com". These initiatives have not always met with great success or public approval; in fact, the last two were picketed a week ago by Republican pundit Pat Robertson and a group of Christian fundamentalists, claiming that they were hothouses of "smut, sin and homosexuality on the Internet".
Clue: Anything with the word "science" in the name, isn't. Computer science is the bastard offspring of two disciplines, primarily mathematics with a little bit of engineering, and is only interesting when a particular problem intersects one of the two. (Cryptography is one example, belonging largely to mathematics.) The whole of the "pure" part of computer science, i.e. the part where it intersects no other discipline, could be completely described in a fairly small paperback book.
An awful lot of "professional" programmers, even those with computer science degrees, hate math. Unfortunately, they also aren't any good at it. That's why we have the current flood of crapware saturating the market. If they'd take the same amount of time working through a system by hand as mathematicians do, and/or designing and implementing it properly the way engineers do, the incidence of security holes and bugs would drop to zero.
A mathematician who sat down and scribbled something out in a couple of days and submitted it to a mathematical journal for review would get laughed out of the community when (not if) a flaw was found. (Hell, after _twenty years_ of hard work, Wiles' proof of FLT had a flaw in it when he released it.) An engineer who sat down and designed a bridge ex vacuo in a couple of months would kill a lot of people, get immediately and successfully sued for incompetence, stripped of whatever professional license he had and would most likely die penniless and broken, particularly if he disregarded the prevailing design methodologies of his profession. Yet programmers routinely churn out crap. (Before you get up in arms, think about how much crappy software there is in the market today. Who do you think writes it?)
Computer science ought rightly to be taught as what it is, a subset of mathematics, rather than as a separate discipline. Programming proper needs to be relegated to what it is - a skill whose mastery is denoted by some form of professional certificate, _not_ a degree - and completely divorced from computer science in general. As taught today, a CS program seems largely to consist of programming language classes (C/C++/Java/Perl/HTML, feh/etc.), which is completely ass-backwards. This among other things needs to change in order to get the programming industry producing quality products.
Slashdot Mirror
Bill Gates
Steve Ballmer
Larry Ellison
Lennart Poettering
Richard M. Stallman
Lucretia Borgia
"Tracking the RNG" would help you win the game, but it doesn't tell you anything about how to play the game.
That would be my point.
This AI learns to play the game, it then wins the game using experience it gains in the same way a human does - feedback from the game score.
That is one possible interpretation, which is not supported by the statements so far. That is not to say that it is not the case, only that it is not currently supported by what I have seen so far; something along the lines of "We tested this against games with multiple RNGs with no perceptible change in AI performance" would support that interpretation. There are other interpretations. People are *assuming* that "wins" = "plays the game" - and the company that did it isn't relieving anybody of that perception (understandably). That's the point. Exploration of other explanations for success are warranted.
Consider that, for games which possess a weak RNG (i.e. predictable starting conditions and knowable changes in game play, i.e. most old console games), it is in theory possible to play *blind* - in other words, not actually paying attention to what's going on on the screen, but simply hitting buttons at precise enough intervals. If 'score' is taken as a proxy for 'how far you can get in the game' (ceteris paribus, someone with a higher score made it longer), then most known machine-learning methods will converge on that/those sequence(s) without any understanding of 'the game' per se. It may even be possible to do that for short gameplay sequences based on pattern matches to known game conditions. While that does get off into the semantic weeds of what 'playing the game' is, it is difficult to differentiate between an AI which has 'learned' to play the game in the sense that it understands abstract rules, interprets game state, and makes decisions about what to do based on that observed state, and a neural network which has converged on the correct list of keystrokes to pwn the computer given certain observed starting conditions. One of them is impressive; the other one isn't, quite so much.
I find myself wondering about the following question:
How did they differentiate "learning to play the game" from "learning how to track the game's RNG"?
Most video games have ridiculously simplistic PRNG generators embedded in them. An AI might get "sidetracked" and learn how to play the underlying RNG output of the game, rather than the game itself. That would yield really good results for most arcade games of this type, I imagine (weak RNG, limited input and timing options, etc.) I don't know if they checked for that possibility.
Easy way to check, though: Reach into the game and substitute a better RNG (cryptographically-strong/hardware/quantum) RNG for the one in the game. That would enable you to quickly determine the difference. If the AI's game performance suddenly goes to shit, it wasn't a real game-playing AI. If it doesn't, well, all hail Skynet, I guess.
"... We wrapped a robot in a dead sparrow and decided to see if we could fool the other sparrows into interacting with our creepy, ghoulish automaton! It's *science*!"
And of course, it was COMPLETELY UNEXPECTED that the grisly abomination stapled to a tree branch triggered aggressive reactions from the other sparrows. Because every living thing JUST LOVES to be confronted with a soulless golem wrapped in the dead flesh of another of its kind. And that never causes pants-shitting terror or anything.
I can see it now:
Sparrow 1: "OH MY GOD! IS THAT... *THING* ... WEARING FRANK'S FACE? IS IT?! FRANK??!?!" ...NOT! ...HIM! IT'S A MACHINE! Help me destroy it! Be his egg-layer one last time!"
Sparrow 2: "It's not him anymore. IT'S!
Sparrow 1: "*snf* OK... OK... oh God, Frank... God help me..."
Yup. Science.
Is there, like, a review board or anything? Maybe that could screen some horror flicks before writing checks for this kind of bullshit? "New rule: If your study is substantially similar to the plot of any one of this library of 100 horror movies, or if it has a plausible chance of producing similar outcomes, we're not going to fund it."
"This would be most dramatic if the intellectual property was produced in one nation under its laws then used without license by another nation to effectively eliminate the benefits of the intellectual property protects."
And what a tragedy *that* would be, huh?
"THINK OF THE CHILDREN!!!!!!"
Yeah, uh, you kind of miss the point here. Rothbard, "Fractional-Reserve Banking" and "Anatomy of the Bank Run", game over, you lose.
The problem with FRB: it constitutes fraud. While I maintain that this statement is prima-facie obvious, I have the feeling that it will escape some portion of the crowd, so we'll try a thought-experiment.
"Fractional reserve" banking says that for every, say, $1000 of demand deposits (ex. checking accounts), the bank need only keep some fraction available at any time, on the theory that not all $1000 will be demanded at once. Yes, that's sort-of true - statistically, most of the time, most people will be content to leave it in the bank, and only call it out as-needed. Most of these calls for money will, in turn, be deposited in another bank, thereby adding an extra level of "protection": inter-bank transfers can be "batched" and resolved on different time scales than the demand-deposit processing (e.g. accounts squared at end-of-day, end-of-week, whatever).
However, what the bank is saying when you deposit money in a demand-deposit account - "your money is available for withdrawal at any time of your choosing" - is, literally, not true. *Your* money has disappeared into someone else's pocket, in that it has been loaned out to some other party as soon as it hits the bank. This is not a "theoretical" untruth - it is a real untruth, in that at all times, the bank is illiquid, i.e. does not possess sufficient capital to redeem all its demand-deposit accounts on actual demand. This is what is referred to in any other instance as "constructive fraud". It is useful to compare this to the eponymous Ponzi scheme, where the fraud consists of there being no actual capital or investment to satisfy the promised payout schedule to current investors, requiring that new investors be found to service existing obligations.
This is not the worst feature of fractional-reserve banking, though. FRB is the gateway for the money multiplier and hence inflation; a bank with a reserve requirement of 5% (larger than the current reserve requirements, note; I believe they're hovering at less than 1%) can, with a deposit of $1000, immediately turn around and loan out $950 of that money. This functionally doubles the amount of money in the economy ($1000 of "fantasy" money, and $950 of "actual" money floating around). That $950 typically gets deposited in *another* fractional-reserve bank, almost invariably with the same reserve ratio (set by the central bank, and reinforced by consumer preferences; ceteris paribus, a higher reserve ratio implies lower interest rates on deposits, providing customers incentive to move their deposits to another bank), at which point the cycle starts again ($950 in "fantasy" money, of which $47.50 is kept on-hand, and the remaining $902.50 lent out). This multiplies that original $1000 of "real" money to something like $20,000 in terms of its real economic effect. SUPRISE INFLATIONSECKS LOL. Then, when you start printing up more money (cue the Fed) and tweaking reserve ratios (cue the Fed again), you wind up with - wait for it - more inflation. Inflation has well-known and universally-observed destructive effects, penalizing saving and encouraging increasing amounts of debt, since debt is paid off with future money that is worth less than the original loan. How's that working out for us so far?
(Side note: it is instructive to note the identity between fiat currency and counterfeit, with the only difference being the identity of the printer of money.)
Austrian economic theory consists of the "duh no-shit" observation that this has an effect on the economy - inflation makes money cheaper to obtain (hey, they print it for nothing!), thereby making marginal enterprises "profitable" under inflationary conditions. This produces the "boom". When those conditions cease - there is an lower bound on the worthlessness of fiat currency, beyond which it is not used except as kindling - all of those bad ideas *come home to roost*, with the acc
Or, what potential patent holders will have to do in twenty years is find their Sonny Bono and get the term of patent rights extended to the life of the Universe plus 99 years.
I wish that people would fucking figure out how this shit works already. Any process that makes use of government to shake somebody down ratchets in the direction of "more shakedown/less freedom" only, until enough people get sick of this state of affairs and begin collecting heads. This is independent of how good an idea you personally think it would be to shake some given person A down to benefit person B. Solution: refrain from shaking people down, don't start in the first place. (And yes, patent and copyright *are* very much shakedowns.)
Hayek. Mises. Rothbard. Read 'em. Learn 'em. Live 'em.
"Make sense" to *whom*?
Neither God nor the Universe are under any obligation to "make sense" at all, much less to you, much less any kind of sense that you would identify as such.
I'm frankly surprised that nobody has mentioned the #1 noise source and probable steganographic message carrier out there: spam. It's ubiquitous, customarily comes with a shitload of SEEMINGLY random strings whose purpose is ostensibly to confuse hash-based and keyword filtering (but which could contain God-knows-what), is easy to do, and doesn't raise any eyebrows. What do most people do with spam? Throw it in the trashcan, of course, they can't hardly get rid of it fast enough. You can scatter it across millions of email address, camouflaging the one you're really sending it to. And only for those with the secret decoder ring would the funny strings have any meaning...
You are making the assumption that "winning" in a market is an absorptive process. Yes, the sequence looks like this:
1) Use larger store of cash to cut all competitors out of the market.
2) Once all other competitors are out of the market, raise prices ("gouge").
3) Profit!
Great Slashdot economic analysis, but you forgot steps 4 and 5:
4a) Watch as old competitors re-enter the market or new competitors arise due to the _demonstrated profit potential of the market_ (remember, I'm gouging, so I'm making Big Bux)
or
4b) Watch as customers, disgusted at high and rising prices, decide they don't want what I'm selling and go find something better to do with their money.
and
5) Go out of business.
Rinse, repeat. The only way to prevent steps 4) and 5) is to ensure that your competitors never arise (using regulations and special privilege backed by deadly force to lock out competitors, aka Microsoft, Ma Bell, USPS), and/or to ensure that customers cannot stop buying your product if they are unhappy with it (i.e. government of all stripes).
Nothing is ever entirely static. People who raise prices in a market make that market more attractive for competitors, not less.
Regulations only serve to prevent the entry of competitors into the market (otherwise, what is the point of a regulation?), allowing existing players to raise their prices - or keep them at a high level - without fear of competition. Sometimes those price raises only cover the actual raise in costs due to the regulation. Mostly they're a lot bigger, since the businessman is guaranteed a lower incidence of competition in his market.
Yeah, isn't regulation great?
Murray Rothbard. Ludwig von Mises. F.A. Hayek. And many others.
Basic truth: Government cannot interfere in a free market - IN ANY WAY - without distorting it. What is the free market? A free market is a market where buyers and sellers are able to meet and make a trade without interference. This trade is mutually beneficial, otherwise it would not have been made. When government interferes in this arrangement, these trades are either not made, or they not as beneficial to both parties as they otherwise would have been. (Otherwise, what would have been the point of the interference? The "best thing" would have happened anyway, making government intervention a total waste.)
You are correct in that government benefits _some_ business. Microsoft, for example, made billions of dollars based on the notional value of its intellectual property, which government secured _at no cost to Microsoft_. You are statutorily unable to copy the software, disassemble or reverse-engineer it, or remove pieces from it to use in your software, without being guilty of a crime. I find it droll that a company which took maximal advantage of this amazing boon now expresses indignance at being charged with "predatory pricing", as though such a thing were possible. (Lie down with dogs, wake up with fleas, etc.) You can examine the effects of government on the market in other industries as well, creating what are commonly known as "cartels".
Enforcement of contracts presupposes a situation where people make and break promises willy-nilly, which _of course_ requires the Hand of God^H^H^HGovernment to correct, forcing everybody to "play by the rules". There is no such epidemic of perfidy about; those people who do not honor their contracts typically discover many fewer takers for their next proposed contract. The problem is self-correcting. Further, even in those instances where a contract must be "enforced", it is a highly dubious proposition that government must be involved; examine the practice of private debt collection as one example, a case where one party made a contract with another and then refused to make good on their obligation. There are a multitude of ways to seek restitution; government intervention is not the only path, and indeed is hardly the best.
There is no such thing as a "pure public good". "Pure public goods" are a misnomer; they inevitably wind up "belonging" to a certain elite class at the continuing expense of the rest of the populace. Something "belongs" to you only if you have complete control over its disposition; if you believe that you "own" any portion of purportedly public property, try taking your aliquot piece of it and using it for your own purposes and see how far that gets you.
Finally, governments do not "bust up" monopolies, they tend to create them. Reread above; Microsoft was created by the government enforcement of copyright of software. If you doubt this analysis, ask yourself this; what are the effects of regulation on an industry? They constitute a statutory (i.e. "non-market-based") barrier to entry to that market thereby creating an unnatural shortage of competition, of course. Monopoly - total control of a particular industry - is a natural extension of regulation. Indeed, once traced back to the root, all known instances of monopoly have sprung forth from government intervention in a market.
You need a lesson or two in economics, I'm afraid. You could also stand a lesson in the precise definition of "anarchy". I daresay you'd most benefit from a heapin' helpin' of Shut The Fuck Up, too, but I have my doubts that you'll partake. Oh well. Have a nice day!
The fundamental problem is this:
The governments of these African countries, like the government of India before them, are in the process of subsidizing the development of what is perceived to be a cash cow of limitless milkability, IT. This process is nothing more and nothing less than seizing money at gunpoint from other, more productive domestic industry (natural resources development as one example) or getting it from dumber countries (like, say, the US and its billions of dollars of foreign aid, ironically likewise looted from the American taxpayer), and giving it to another industry to make it grow in defiance of market forces. Governments are subsidizing the production of millions of PhDs, handing out favors to "tech-savvy" "entrepreneurs" and foreign companies to take advantage of the perceived riches of the tech industry, not realizing a couple of very basic tenets of economics:
ALL OTHER THINGS BEING EQUAL, WHEN SUPPLY GOES UP, PROFITS (AND PRICES) GO DOWN.
and
IT IS NOT POSSIBLE TO CIRCUMVENT MARKET FORCES. USING GOVERNMENT TO FORCE THE ISSUE LOOKS BETTER NOW, BUT COSTS MORE LATER.
The problem is, this is not an endless phenomenon. It wasn't profitable to locate things in India before, for a multitude of reasons (lack of infrastructure, lack of education, social problems, whatever). It will likewise be unprofitable in the future, when their millions of PhDs are hacking cabs in New Delhi to make rent, or becoming farmers. (You can see this process beginning now. The market there has reached capacity, and other places - like Africa, Land of Ceaseless Warfare, Spam, and Disease - are being seriously considered as places to invest in tech, because the market in India is getting too inflated.) It sure as hell has been unprofitable and/or just plain dumb to locate any form of tech industry capital in basically any African country, where the odds of its being nationalized, destroyed, or devalued in the customary and predictable political upheaval are astronomical.
The cornucopia of benefit from IT and tech in general is mostly illusory. It came about in the US largely through a government/Federal Reserve easy-credit policy in the 90s that allowed all manner of idiocy to get funding and look great on paper (AKA the dot-com boom - pets.com, anyone?), followed by the bust when all of these crappy investments based on bullshit were exposed as the stupid ideas that they were. Yes, there is some benefit to tech, as long as it enhances productivity and quality of life. No, its benefit on life and productivity are not infinite, nor is this benefit anywhere near as bountiful as some think. It seems that the governments of other countries, enthralled by the idea of a trillion-dollar business tax base (or "loot pond") springing up overnight with a minimum of effort, are going to go down this same road with precisely the same heartbreak at its end. The citizens of these countries would do better to leave their neighbors alone and spend their time farming and defending their property from invaders. After a few decades of respect of property rights and natural rights have set in, then they could begin working their way up the industrial/informational ladder, and would be in a much better positioin than we are now. (For that matter, we in the US should probably take the same advice.)
Oh well.
NB: This list was lifted from http://www.craphound.com/spamsolutions.txt I am uncertain of further attribution.
... Which we don't have now, and for various reasons, will probably _never_ have. Why not predicate the existence of a functional antispam system on something vaguely probable, like a secret group of four superhumans wearing leotards? We could call them the "Spamtastic 4", and they could fly around preventing evil spammers from spamming. Ooo, and they could have a moon base too. Yeah, that's the ticket.
Anyway.
You're missing the point, which is entirely usual for an AC. The response in the FAQ/QnA to all questions of "How will we handle X?", where X is "unbonded/poor/anonymous/whatever senders", "mailing lists", etc. have the same two answers: "Well, you'll have to put them on your whitelist", or "you'll have to look at the messages individually to determine if you want to read them".
BUT WE DO THAT NOW. There is nothing new here, except an elaborately-designed bit of wankery designed to flag potential spam messages in a slightly novel way. You still have to undertake the Aegean task of discovering which ones are spam and which are not, unless you are willing to forego communication with everybody who is not on your whitelist and/or who does not post a sufficiently large bond. The very best you can hope for from this system is to recover some amount of money for reading spam, THEORETICALLY. (The ability of this service to guarantee this has not been demonstrated to my satisfaction, any more than the existing system's capability to prevent spammers from obtaining access in the first place.)
To address some points of your "rebuttal":
"(x) Mailing lists and other legitimate email uses would be affected
Incorrect, see other posts"
No, not incorrect. It increases the difficulty of running a mailing list, to say nothing of increasing its costs.
"(x) Users of email will not put up with it
Depends on the effectiveness and the cost. This system promises legitimate users negative cost!"
Hell, I can go through my spam archive and find messages promising me wealth, a 50-foot penis, and hot sex with teen virgins. I don't believe that, either. (Wait until that legitimate user gets every bond redeemed due to losing the eternal online popularity contest, or even better, has his bond account emptied due to a security flaw somewhere. We'll see what he/she thinks of it then.)
"(x) Requires immediate total cooperation from everybody at once
Incorrect, a non-participating recipient will simply not request bonds. As with most other anti-spam solution, a fall back address can be used which is checked with lower priority and stricter content rules to discourage users from sending mail to the non-participating address."
Thereby providing, for 99% of all email users, zero benefit over the existing system. (A "solution" where you still have to clean a spam trap is not a solution.)
"(x) Armies of worm riddled broadband-connected Windows boxes
Correct, a working and secure micropayment system is a requirement."
"(x) Joe jobs and/or identity theft
Incorrect. Why would you accept to provide bonds for messages which you didn't send?"
I wasn't aware that you had to agree to an identity thief/cracker emptying your bond account of however much you had in it. A security flaw hardly requires your permission to exploit. (But now that I know you have to agree to it, that makes it much better. Tell me, do the attacks against the system require the use of the RFC 3514 "Evil Bit"?) For extra crunchy added badness, ponder possibilities like bond accounts being attached to checking accounts, and the average state of home network security. ("Hey, I'm gonna go wardriving and spam myself a new Mercedes!")
"(x) Bandwidth costs that are unaffected by client filtering
Incorrect. With increasing popularity of Atte
(As a side note, what happens if you receive mail without an associated bond? 12.2Q in the Q&A says "Well, you could still read it", which OBVIATES THE ENTIRE FUCKING POINT!!! Yet another idiotic spam "solution", in other words. Oh well. Here's where it scores on the Spam Solution Checklist:)
Your post advocates a
( ) technical ( ) legislative (x) market-based ( ) vigilante
approach to fighting spam. Your idea will not work. Here is why it won't work. (One or more of the following may apply to your particular idea, and it may have other flaws which used to vary from state to state before a bad federal law was passed.)
( ) Spammers can easily use it to harvest email addresses
(x) Mailing lists and other legitimate email uses would be affected
( ) No one will be able to find the guy or collect the money
( ) It is defenseless against brute force attacks
( ) It will stop spam for two weeks and then we'll be stuck with it
(x) Users of email will not put up with it
(x) Microsoft will not put up with it
( ) The police will not put up with it
(x) Requires too much cooperation from spammers
(x) Requires immediate total cooperation from everybody at once
(x) Many email users cannot afford to lose business or alienate potential employers
( ) Spammers don't care about invalid addresses in their lists
( ) Anyone could anonymously destroy anyone else's career or business
Specifically, your plan fails to account for
( ) Laws expressly prohibiting it
( ) Lack of centrally controlling authority for email
( ) Open relays in foreign countries
( ) Ease of searching tiny alphanumeric address space of all email addresses
(x) Asshats
( ) Jurisdictional problems
(x) Unpopularity of weird new taxes
( ) Public reluctance to accept weird new forms of money
( ) Huge existing software investment in SMTP
( ) Susceptibility of protocols other than SMTP to attack
( ) Willingness of users to install OS patches received by email
(x) Armies of worm riddled broadband-connected Windows boxes
( ) Eternal arms race involved in all filtering approaches
(x) Extreme profitability of spam
(x) Joe jobs and/or identity theft
( ) Technically illiterate politicians
(x) Extreme stupidity on the part of people who do business with spammers
( ) Dishonesty on the part of spammers themselves
(x) Bandwidth costs that are unaffected by client filtering
( ) Outlook
and the following philosophical objections may also apply:
(x) Ideas similar to yours are easy to come up with, yet none have ever
been shown practical
( ) Any scheme based on opt-out is unacceptable
( ) SMTP headers should not be the subject of legislation
(x) Blacklists suck
(x) Whitelists suck
( ) We should be able to talk about Viagra without being censored
( ) Countermeasures should not involve wire fraud or credit card fraud
( ) Countermeasures should not involve sabotage of public networks
(x) Countermeasures must work if phased in gradually
(x) Sending email should be free
( ) Why should we have to trust you and your servers?
( ) Incompatiblity with open source or open source licenses
( ) Feel-good measures do nothing to solve the problem
( ) Temporary/one-time email addresses are cumbersome
( ) I don't want the government reading my email
( ) Killing them that way is not slow and painful enough
Furthermore, this is what I think about you:
( ) Sorry dude, but I don't think it would work.
(x) This is a stupid idea, and you're a stupid person for suggesting it.
( ) Nice try, assh0le! I'm going to find out where you live and burn your
house down!
Great. So AMD has apparently rediscovered the Harvard architecture, or a variant thereof. (The "virtual" Harvard architecture?) [ http://en.wikipedia.org/wiki/Harvard_architecture ]
Has it been adequately demonstrated yet that there is in fact nothing new under the sun? No? Okay.
For a bonus laugh, I wonder if they're going to get a patent on this "novel technology".
Hmm. Three stories down, it appears as though Intel has a demonstration of "superfast communications technology in excess of 2 Gb/s" next week.
Coincidence? I think not.
Sounds like someone at Intel forgot to do their mental Kegel exercises. "It's done! It's amazing! It lives!" "Umm, wait a minute. How can we actually _demonstrate_ this?"
I suppose turning the convention hall into a 32,000x32,000 full-surround-sound 60-fps digital pornograph would be singularly unwelcome at the developer's conference, but... ah, wait. Never mind. (Only problem is, do they have enough time to do it? And do I have enough time to attend? Problems, problems...)
First of all, it would be "libel" if it is published in a written form. "Slander" entails _verbal_ communication. Second, truth is an affirmative defense for a charge of libel. What does that mean? It means that if what was said in the article was _provably true_, then it doesn't matter if there was any hostile or malicious intent on the part of the author. You cannot be fried for reporting fact. If "TheDeacon" has a foot fetish, and Joe Reporter reports that he has a foot fetish, whether or not he portrays that in a favorable or unfavorable light, it is _provably true_ and therefore is not libel. Third, being a private citizen is not a silver bullet to prevent libel. Yes, it is true that being a public citizen makes it nearly impossible to successfully sue for libel. That does _not_ mean that being a private citizen makes it _easy_ to successfully sue for libel. Someone else posted the three criteria for defining "libel" above, which must _all be met_, and which I encourage the reader to go back and review. I will concede that IANAL, but unless my memory core is getting seriously rusty these days, I know these preceding three points to be true nonetheless.
Lose the FUDstorm and pick up a a copy of the Reporter's Stylebook, Slashdot. And find a fucking dictionary while you're at it. It contains words and their _proper_ meanings.
Surprise - that's not what I said.
What I _said_ was that this is not the new Big Iron, because a cluster of PCs does not qualify. It is not "engineered" in the sense that a Cray is, and I'm willing to bet that if you ask an actual Cray engineer you'll get your ass kicked for implying that some bozo on the street with a half a brain and a few bucks can build a supercomputer that will beat out something designed by a highly-trained engineering team from one of if not the best supercomputing manufacturers and research corporations.
To address the next point, nothing about the Flat Neighborhood Network makes it suitable for finely-grained problems, or at least as suitable as a "real" supercomputer. Throwing more switches at the problem and being clever with the wiring helps, but not to the extent that it will beat out a well-engineered chunk of Big Iron. Perhaps most importantly, there is a limit on the design of this network imposed by mathematics; discovering the proper "wiring" of the network is a nasty combinatorial problem, and is (according to them) difficult to solve for even 64 processors. Try doing it for 128 and observe the crispy crunchiness of this problem as you gouge out your own eyeballs and go insane. This type of cluster is not fundamentally different from a Beowulf cluster; while the result may well be useful for a certain class of problems, it's not worth getting all gooey about, primarily because there's _nothing new going on here_. Christ, if you want a big supercomputer, look at distributed.net. I bet they'll be able to beat 10 TFLOPS, at least theoretically, and d.n is yesterday's news. It just flat-out doesn't lend itself well to some problems of great interest. Yeah, it'll crack batches of keys just as slick as you please. What it won't do is something like weather modeling, at least not without bringing the Internet to a screeching halt as all available bandwidth is saturated with data requests.
Looking at the Cray site, they certainly are offering a cluster supercomputer. They're also offering several other types of supercomputer as well. As I said before - a Beowulf cluster is not the solution to all problems. Show us the benchmark on this bad boy for something nice and nasty, like turbulence modeling. I bet that all those processors choke nice and hard on the bottleneck and drop that 10 TFLOPS figure to something much more realistic, like about 1 TFLOP, in this circumstance.
-SD
The title says it all. Big Iron is _engineered_. No matter how big or how spiffy a Beowulf cluster is, it's still just a bunch of PC motherboards kludged together with a bunch of network cards. There is a reason Crays are expensive - they are _worth it_ from a performance standpoint, because not every problem lends itself easily to the solution of a Beowulf cluster. Some problems require the exchange of a lot of data between a lot of nodes, and a little math will show that it won't take much data interchange to saturate even a GigE switch. Adding more machines is not going to help; craftily designing and overengineering the network _might_, but by the time you get this whole damned thing glued together well enough to approximate a Cray's performance, you'll have spent enough to have just flat-out bought a Cray in the first place.
As others have noted, while this thing may have a theoretical peak performance of 10 TFLOPS, I'm willing to bet that number goes down like Monica Lewinsky on Quaaludes when you feed this magical supercomputer a problem that's _not_ suitable for distributed.net (i.e. one where computations on one node are dependent on computations on another node, like fluid-dynamics problems, turbulence, etc.)
Yeah, it's interesting as a curiosity, but this is by no means spectacular. Beowulf is good for what it's good for, which is a "poor-man's supercomputer" that works well for coarsely-parallel problems that don't require a lot of internode communication. It's not the Philosopher's Stone, folks.
-SD
The problem is that current advances in computing technology are focussing on "infrastructure"-level improvements; improvements in the actual hardware. A couple of companies (Apple) have spent money on the user interface. Noone, to my knowledge, has made a credible or successful effort to construct a metadata layer above commonly used information, indexing and sorting it so it is presented in the most effective fashion and linked to things that make sense. Notes:
a) I am _not_ talking about a language to do this in, because I'm sure some nitwit will chime in with "What about (XML|whatever TLA is the newest buzzword)?" That is "infrastructure". I am talking about actually _doing_ it, which hasn't been done in any global context yet.
b) Think Star Trek. Ever notice the funky engineering consoles that changed menus and key positions based on whatever information was being accessed? Playing with the warp-core antimatter injectors, or whatever, popped up a different screen and a different means of playing with it than firing the phasers. This is analogous to the process necessary to render information "usable"; Mr. Gelernter, although obviously floating off in a nether zone all his own, nevertheless gets this one right. People have to sit down and engineer default "pathways" through the morass of information out there with lavish effort and much cursing, not to mention a healthy dose of voodoo. The infrastructure to do this has to be ubiquitous, reliable, and standardized (more or less). Bill Gates' quote seems to be taken out of context; I think he's attempting to say that there are no intelligently designed links between most pieces of data that are stored on the average PC, and this is something that needs to change. As it so happens, he's right on that score.
$0.02 for free, it's worth what you paid for it.
-SD
This is not a new idea, and the author of this "whitepaper" is a twit.
Some time back on a mailing list far, far away, this got knocked around by myself and a couple of other people. (This was around the time of CR2.)
As I conjectured at the time, and still believe, the primary reason most worms fail is that they are either
a) written in a language with which 99% of programmers have very little skill (C, C++), or
b) written in a remedial computer "language" that is useless to begin with (VB, macros, etc.)
This is true with every known worm to date. The problem with VB and languages of that type, well, never mind. If you can't figure it out yourself, stop reading now. The problem with C and C++ is that it is extremely easy to shoot yourself in the dick with either of these languages. (See the Morris worm's replication-rate bug for an example.) The problem is that with such a language, if you do not do everything exactly right, there will be a flaw in your worm and it will most likely fail, subtly if not spectacularly.
I conjectured that it would be a Good Thing (for he who wants to 0wn a large portion of the Internet) to investigate other languages instead. Prolog came immediately to mind. It seems ideally suited for this task as it is an intrinsically goal-oriented language. In addition, all the "complexity" of a Prolog program is hidden inside the resolution engine (which can be made quite small), so given that the resolution engine is operating correctly, if your program is well-designed, you have very few surprises to worry about. (Typically, although this is by far not a globally true statement, Prolog programs with bugs in them simply don't work, rather than failing subtly while appearing to work. That seems to make them a helluva lot easier to check.)
The fact that Prolog is an interpreted language makes it nice and portable; it is possible to create system-dependent and system-independent goals, and let the program figure out where it is and how to deal with it at runtime.
Also discussed for this bit of nastiness (I call it an "amoeba", after the slime mold which has similar traits) were the following:
a) goals added to the database from non-local sources would require a signature for each goal or batch thereof to prevent poison goals from being automatically propagated through the network (said signature using a private key, held by the Evil Overlord, matching a public key distributed with the worm);
b) "local" can be defined as being inside a clique of some finite size, say 10 computers, where goals and derived goals are propagated freely between members of the same clique, but not from members of different cliques unless they bear a valid signature;
c) infection can be maintained within a clique and progress to new machines from the "axolotl" effect; that is, machines will attempt to replace hosts in the clique that are no longer communicating, and at some random point, will decide all other hosts in the clique are "dead" and will seek to regrow the entire thing;
d) communication is probabilistic, not deterministic, since this is an effective means of frustrating attempts to locate infected machines; communications attempts are ignored at random, regardless of source or content, and spurious communications will be sent (random recipients, forged IPs, etc.)
e) Randomizing of opcodes using several different tricks, like instruction reordering a la Intel, instruction rewriting (essentially, replacing instructions with a set of different instructions that performs the same task), etc. The rewriting bit is essentially lambda calculus. >;-> Statistics for rewriting seem to be fairly nasty; I haven't had any luck coming up with an analytic expression for them, and have foregone a mathematical approach for a good old-fashioned "high-water/low-water" method for dealing with the fact that unless all isomorphic instruction strings are the same size (which they're not), this thing will have some serious problems with unbounded growth at some point.
These are a few of the ideas involved in this whole thing. Note carefully that none of this stuff is revolutionary; rather, it simply brings some old concepts back and puts them in a newer context. Once someone jerks their head out of their arse and figures out that C/C++/Java/whatever are not the best languages to do abso-fucking-lutely-everything in, this sort of thing will pop up and will be nastily difficult to defeat. (If indeed it's even discovered.)
-SD
I dunno. Waiting for a lengthy fsck can be very satisfying at times. Other times, I just want to get a quick fsck and go on about my business. (Typically around noon.) Depends on the day, I guess.
The advantages to such a system? Well, like you said, if you can afford to wait for a lengthy fsck once in a while, you're okay. If you just can't contain yourself, though, it might be worthwhile to go ahead and get it over with. The only problem with this is that if you absolutely cannot make it all the way through a slow fsck, you might pick up a reputation and/or an unflattering nickname ("Minuteman", for example.) It all depends on what you want; if instant gratification is important to you, you should be fine, modulo the occasional heckling and snickering by coworkers who are plugged into the office rumor mill. (But see touch(3); it may be more worth your while if you're all about speed as opposed to endurance, and might help with the problem of other people's cruel tendencies.)
Hope this helps.
-SD
As though 802.11 wasn't bad enough. Now we can have someone sniffing hard drive accesses as well?
I wonder when "Bluesnort" will be coming out. >;->
Does anyone know if the encryption for Bluetooth is as braindamaged as some of the others out there at the moment, or if it's actually something halfway decent?
-SD
the Chief Rabbinate of Israel, wanting to jump on the online-learning bandwagon, has designed a new curriculum. The University of Tel Aviv now offers a 90-day online course in circumcision, titled "Who wants to be a Mohel?"
Reception from Jewish communities has been lukewarm at best, especially as a typo in the online documentation for the course recently led to some accidents with unskilled graduates mistakenly "collecting tips", unaware that this refers to the custom of providing a gratuity for the mohel's services. Embarrassed by this, the leadership of the University of Tel Aviv has temporarily suspended the curriculum "for review".
This follows a rash of online-learning initiatives from respected universities, including Oxford's "eParasitology", Cornell University's "RealProctology", and Johns Hopkins "myGynecologist.com". These initiatives have not always met with great success or public approval; in fact, the last two were picketed a week ago by Republican pundit Pat Robertson and a group of Christian fundamentalists, claiming that they were hothouses of "smut, sin and homosexuality on the Internet".
From FakeReuters.com
-SD (Note - this is all a parody. >;->)
Clue: Anything with the word "science" in the name, isn't. Computer science is the bastard offspring of two disciplines, primarily mathematics with a little bit of engineering, and is only interesting when a particular problem intersects one of the two. (Cryptography is one example, belonging largely to mathematics.) The whole of the "pure" part of computer science, i.e. the part where it intersects no other discipline, could be completely described in a fairly small paperback book.
An awful lot of "professional" programmers, even those with computer science degrees, hate math. Unfortunately, they also aren't any good at it. That's why we have the current flood of crapware saturating the market. If they'd take the same amount of time working through a system by hand as mathematicians do, and/or designing and implementing it properly the way engineers do, the incidence of security holes and bugs would drop to zero.
A mathematician who sat down and scribbled something out in a couple of days and submitted it to a mathematical journal for review would get laughed out of the community when (not if) a flaw was found. (Hell, after _twenty years_ of hard work, Wiles' proof of FLT had a flaw in it when he released it.) An engineer who sat down and designed a bridge ex vacuo in a couple of months would kill a lot of people, get immediately and successfully sued for incompetence, stripped of whatever professional license he had and would most likely die penniless and broken, particularly if he disregarded the prevailing design methodologies of his profession. Yet programmers routinely churn out crap. (Before you get up in arms, think about how much crappy software there is in the market today. Who do you think writes it?)
Computer science ought rightly to be taught as what it is, a subset of mathematics, rather than as a separate discipline. Programming proper needs to be relegated to what it is - a skill whose mastery is denoted by some form of professional certificate, _not_ a degree - and completely divorced from computer science in general. As taught today, a CS program seems largely to consist of programming language classes (C/C++/Java/Perl/HTML, feh/etc.), which is completely ass-backwards. This among other things needs to change in order to get the programming industry producing quality products.
-SD