SCO has also demanded the BBC turn over working blueprints of the TARDIS, that Arthur C Clarke provide a CVS snapshot of HAL 9000 and that Isaac Asimov reveal the equations used in psychohistory. SCO is also investigating secret codes in the Bible for possible hidden prophecies revealing System V code.
I am now convinced that someone at SCO has flipped their lid and become a paranoid schizophrenic. Either that, or they are aiming at a career on the Comedy Channel once SCO sinks without trace.
They get it that people are suckers and will pay twice for the same product (once to the manufacturers in higher costs to pay for the ads which pay the cost of the broadcast, then once to the broadcasters directly for the on-demand).
They get it that they can scam, bilk and price-gouge legally, so long as it is in small amounts at a time.
They get it that they're going to be able to milk the compliant viewer for all they're worth (and then some), boosting profits and keeping the shareholders happy, while keeping the programs sub-standard and their integrity non-existant.
From the perspective of the TV companys, what is there not to get?
TV stations were being relayed over CU-SeeMe and the Multibone over ten years ago. Legally. Without advertising. Free. This "new stuff" is old-hat - or would have been if users had pushed for the technology to be widespread. As it was, CU-SeeMe denegrated into porn and commercialism, then vanished into the abyss. The multibone persists, but virtually no ISP is being encouraged to carry it.
For a website, MySQL is probably the superior option. There is (almost) always a tradeoff between functionality and performance, so it is (usually) wisest to go for the option that is closest to what you want, as that will perform the best.
If the site is very small and the database code is not particularly advanced, you might even do better with SQLite. However, this does require using PHP 5. If you're on PHP 4 or earlier, that's not an option.
Now, this changes dramatically if you need very advanced database features. Postgres is certainly far more advanced than MySQL (although the gap is not as great as it once was). It can also scale better, but for a small site that probably won't matter so much. A website that allows you to manipulate geographical data could be small/hobbyist but would still require an awesome database to drive it.
The word "wheelbarrow" originates from the ancient burial mounds called barrows and are traditionally the same shape as the more ancient of these, called "long barrows". The common theme being dead stuff buried under large mounds of earth.
Databases are often used for dead projects, buried under large mounds of data. The parallel would seem to be pretty exact.
Open Source as a business model is looking like it could do something similar to the.Com era. Ingres has a reputation in Big Business that PostgreSQL and MySQL do not (fair or unfair). If (as is speculated) Open Source goes through a major boom, Ingres will be hot property with database developers and therefore potentially offer a good return on the investment.
(No, this company isn't likely to hold onto it. This looks like the sort of deal where they buy something that might be worth a lot more next week or next year, so they can sell it and make money. Simple speculation.)
Note that this theory ONLY holds up if the price was sufficiently low that if Open Source does have a boom, the investors will be able to make significantly more profit on it than they would have done speculating on, say, stocks or currency. So if it's a land-grab, they're expecting the value of Open Source software to at least double or triple in the next few years. Otherwise, the investment wouldn't be worth it.
(The value of Ingres is important in this - if it cost them $10, then a 100% profit margin won't be hard to achieve. The more they actually paid - never mind anything officially said - the more it will take to make the profit margins they'd have to be looking at.)
There is one other possibility, but it's a remote one. Instead of a boom, they're expecting a crash - particularly in the database market. They'll have IT guys and the Open Source mantra of being able to look at the code can be persuasive. They may be convinced that the database market is so glutted and so unsure of direction that it will have to collapse. If it does so, then the only support they'll have is whatever they give themselves anyway.
In this scenario, they're playing the role of survivalists - acquiring the technology they'll need to survive, expecting things to get nasty for everyone else (ie: everyone still on their feet playing the role of SCO). In this scenario, the only sure defence is to have something with no prior technology in it - and Ingres would meet that. This does seem unlikely, but I wouldn't rule it out without a lot more information.
SQLite is good if you want an embeddable SQL engine, but it isn't exactly a heavy-duty workhorse. To serve dynamic content for a single-user website, where Big Iron Databases would look stupid and do nothing extra, this is exactly what you want.
If you want something that'll do a bit more, but retain a lot of the speed and also keep the footprint down, MySQL is probably the best bet. It has a lot of the functionality of the really large databases - perhaps rather more than is good for a lean, mean database machine - but still gives good balance between function and performance.
PostgreSQL is much more powerful than MySQL, but at the cost of being bigger and generally not as nimble. If you're dealing with mid-sized Enterprise-level databases, I frankly wouldn't care about per-transaction performance as much as I'd care about maintaining good performance for respectable databases for a few hundred simultaneous users. MySQL would be hammered long before PostgreSQL for this kind of work. For GIS stuff, where you've some fairly tricky topographical information to manipulate, you'll notice that PostgreSQL has a far bigger following.
This leaves Ingres, which has a reputation for being good for Big Corporate Data Warehousing. Data warehousing is a very different problem from regular relational database handling. The problem-space is typically orders of magnitude greater, the database generally isn't going to be normalized and the mappings are generally altered to be less I/O-bound - which usually means more work for the CPU.
The problem-spaces solved by these databases are all very very different. I would love to see a database that had pluggable components such that different components were optimized for different types of workload and that different functions could be loaded/unloaded as needed, so your footprint was always the lowest for what you were doing, not what the database was capable of.
The fact is, no such database is in wide use - assuming it exists at all.
And we're only talking about the SQL Relational model. The "pure" relational model (as discussed on Slashdot many times) is different again, as are the Object-Relational and the Object-Oriented models. Absolutely none of these could be used with Xanadu's ZigZag data structures (the relationships are essentially permutations and order-independent, whereas formal databases use relationships that are either one-way or follow a specific ordering). They generally don't distribute well, either, as data is bulkier than code, forcing you to load-balance rather than distribute logically.
(That last part can be solved for some cases - it is fairly common to have a mix of tightly-coupled and loosely-coupled data, so it is possible to split the problem-space in a way that keeps communications down and takes advantage of a parallel architecture. It just isn't easy.)
Ingres is important as a database, because PostgreSQL can't (yet) cut it in the Really Big Database World. As good as PostgreSQL is, I would not want to replace extremely large-scale Informix or DB/2 databases with it. Maybe someday, but not today. Ingres - I'd give it a maybe. It does have the reputation of being able to handle it.
The multitude of engines out there are also important, because there are many, many different problem-spaces out there and NONE of them - not one - is good at even a few, let alone many, and certainly not all. And there are many problem-spaces for which there are no databases at all. At least, none worth mentioning.
This is a fixable problem, but not until someone goes out there and fixes it. That isn't happening. So, until then, I'll use RRD, SQLite, MySQL, PostgreSQL, Ingres, Sleepycat DB4, ZopeDB, OpenLDAP, CUPS*, Shadow*, Postfix*, Reiser4*,...
*These all have databases in them. The password file, the print spooler, all the fancy file-access features of Reiser4... And because they're all working in subtly different ways, you WILL have a database engine for each, until or unless someone produces a system that can do all of these as well as each of the specialized solutions.
They get everywhere. Even when you're minding your own business in the bloodstream of a duck, they'll STILL manage to track you down and take pictures with those really powerful telephoto lenses.
This should terrify celebrities the world over - if something this small and elusive can't escape the mercenary photographers of the tabloid press, what can?
Virtually everything written by Fermat was also wrong. However, Fermat's Last Theorum has ended up not only being right (in a way Fermat could not possibly have known), but has shown a fascinating relationship between two different geometrical systems.
Does this mean we should take every crackpot seriously? No, of course not. What it does mean, though, is that serious scientists should be encouraged to not rule something as meaningless merely because it is likely incorrect as stated. If we'd done that, we'd never have learned anything about anything.
A trivial example is the "Cold Fusion" fiasco from Fleich and Pons (spelling may vary) from Utah. Well, I doubt anyone seriously expected anyone from Utah - especially chemists - to stumble onto anything interesting in physics. And, surprise surprise, they didn't. What they DID stumble onto, however, was a very interesting form of fuel cell that can store fairly large amounts of hydrogen within the cell.
ObTrivia: The problem with Apollo 13 was that hydrogen and oxygen stored for use by the fuel cells was vented into space after an explosion. Conclusion: If the fuel was stored in a chemically stable form, which could be electrically released to generate more power than was used to release the fuel, then you'd have an fairly accident-proof fuel cell. If the fuel was then contained wholly in the cells, you would need no fuel tank or fuel lines, removing a problem with existing hydrogen technology.
Can these claims have any meaningful value? I don't know, but I do know that if they do and they are 100% ignored because they're meaningless as is, we never will know. The trick is to learn what is useful without being burned by the useless, discarding that which cannot be usefully learned from without discarding information which would save time to examine closer.
Algorithms such as AES will be broken by flaws in the algorithm - there are already some areas considered a little suspect, I believe. I think that brute-force methods will continue to evolve, but in tandem with such weaknesses. You absolutely need to reduce the search space to bring the number of keys to be searched to a managable level.
Now, I can see key lengths increasing - hashes are up to 512 bits, so 512-bit keys would seem a logical step. NIST are researching encryption modes that provide a much higher level of security, and this is another area I see getting a lot of attention in the future.
So encryption is definitely not a dead subject and I think AES will be seen as naively weak in the sort of timescale you're giving (a century or so). Even Serpent (another AES contender) is only rated as secure for another 50 years unless the algorithm has been broken before then.
Multi-pass encryption with multi-pass modes that are tamper-resistant and repudiation-resistant would seem the next logical step in encryption technology. Multi-pass is good, because encryption algorithms don't randomize sufficiently and it is often possible to extract some contextual information.
I also think it likely we'll move away from symmetric ciphers to asymmetric, provided a good parallel algorithm can be found. CPU cycles are cheap, these days, so the old excuse that public key encryption was slow is no longer so valid. A solid parallel algorithm would demolish that reason altogether.
First, the British police probably can't use the NSA to do decryption work. Second, unless the NSA has actually broken the algorithm, they are simply not going to be capable of breaking any code longer than about 64 bits in the useful lifetime of the data. If the encryption algorithm is a one-time pad (or a close faccimilie thereof), then they will be incapable of breaking it at all without the key. OTPs are, quite literally, unbreakable because EVERY possible decryption is equally likely.
Is hopelessly optimistic. Let us say you had a processor capable of a billion (2^30, not 10^9) operations per second and that you've hard-coded the processor such that you can try one key in one operation. You can now break a key of 30 bits in 1 second. Let us also say you've built a large grid computer with 1024 nodes in it, so you can do one trillion (2^40) keys per second.
Such a computer can break an ordinary (56-bit) DES key in 18 hours, 12 minutes and 16 seconds at worst. The average time to break a DES key on such a machine would be 9 hours, 6 minutes and 8 seconds.
To break a 128-bit key would require the computer to run for 2^88 seconds, or 9,813,705,283,528,192,184 years.
I believe it to be possible for something to be truly, totally original - for an idea or artifact to have no precursors in part or in whole. If that were to be the definition of innovative, then there would have been perhaps a dozen true innovations in the entire history of humanity. I also believe that - because such work is so rare - that such innovations should be prized above anything else on Earth. There is absolutely nothing in the Universe rarer than creative talent at this level of purity.
If you're willing to relax the rules a little - require no obvious precursors - then the number of innovations grows considerably. But they're still pretty thin on the ground. I'd say that there would be maybe one or two such innovations every decade, somewhere on the planet.
If you go further in relaxing the definition, and merely require no trivial precursors (ie: it cannot be a simple extrapolation or interpolation of what already exists), then there are maybe half a dozen each year.
I tend to go somewhere between those last two cases, in the way I look at innovation. I want the term to be used often enough to be meaningful to people, without being so broad as to be meaningless within itself.
However, I would like a clear term that is expressly for the rare cases that meet the first definition. A spectrum isn't just defined in terms of the progression along it, it is also defined in terms of the extreme ends. Calling them "totally unoriginal" and "totally original" tells you nothing other than they're points on the originality continuum. Zero and infinity, on the number line, are not just points. The expression of zero unleashed whole areas of mathematics, as did the expression of infinity. Their value in defining the limits vastly exceeded the value of any other point on the number line.
What this argument over innovation does is tell us about the line of creativity. It does not define the limits and we're all still arguing over the relative merits over some small segment of the line in its entirity. I think it's time to study the line, the dynamics of it, the maths of it, and the limits of it. Do that and we won't just agree on what is innovative, we'll be able to express the innovation in a way that has meaning.
Oh, I'll agree with that, certainly. And if you use the definition of "innovative" that I've suggested elsewhere (ie: that it has a non-obvious, non-derivative creative element that extends without being an extrapolation and is integral without being an interpolation), then Unix would be innovative.
It's still a derivative, in the sense that it is not wholly original, but it is not wholly derivative, it has innovative aspects that were ingenius and original. To me, it is one of the clearest cases of where you can have it both ways.
The first implementation is indeed innovative. I'd call Gopher and WAIS truly innovative as information delivery systems and the web as innovative as a presentation system. For plain information, Gopher and WAIS are still technically superior.
I would argue, however, that later generations can be innovative - provided they do something revolutionary in and of itself. For example, CERN's webserver was the first (and therefore innovative by being first) but I'd consider NCSA's webserver to to have posessed qualities that the CERN server did not have, in a manner such that NCSA's webserver (IMHO) deserves the title of innovative as well. Although Apache has yet more qualities, I would not consider those to be in a manner that justifies such a title.
By implication, I'm saying that a quality must have some attribute that distinguishes itself above and beyond being a mere addition, for the idea/project to be called a true innovation. To me, that attribute is that the addition not simply be an extrapolation or an interpolation of what already existed but must exist outside of the covered space, yet intersect the covered space in such a manner that the extension is a natural extension, not forced.
The "dumb person's test" for true innovation is that it must be so difficult to see in advance that it had truly occured to nobody at all. EVER. Yet be so obvious once found/developed that nobody really realizes it hasn't always been there,
UNIX was a vastly simplified, cut-down version of MULTICS - even the name is a derivative. If I remember my comp sci history, MULTICS was based - to a degree - on TOPS and Project MAC. I can't tell you if either of those was novel or not.
There are a number of interesting research OS' out there, but I couldn't tell you if any of them were derivative or not. uMicro's Object Oriented OS in the late 80s was pretty interesting. Slow, limted but interesting.
The Transputer/Occam coupling was intriguing in that it provided a layer that wasn't strictly an OS, but provided the system calls an OS would have. As such, it could be considered totally non-derivative, as there was nothing there to be derived. I miss the Transputer - it was by far the best design of processor for its day, and its fate was a miserable, undeserved one.
Cone speakers are lousy for extremely low frequencies. Because of the way you generate a pulse (switching the sense of an electromagnet to vibrate a permanent magnet), the system isn't going to work well at very low speed. As other posters have noted, it will also cause problems because you're trying to move very small amounts of air at a time. Sound falls off with the square of the distance at best, but because of the laws of conservation of momentum and energy, I would be surprised if extremely low-energy mechanical signals were able to reach that kind of efficiency.
What you need to do is move a far, far larger column of air. There has been work on using electrical sparks to generate pulses of sound which you can then modulate. This is not mechanical, so you get better efficiency on converting to sound. It also has far better response, would be capable of handling larger amounts of energy, and the weather systems have been demonstrating a very large-scale subwoofer for billions of years.
In all seriousness, I do believe that if you built your own "lightning generator", with diaphram, on the order of perhaps a few hundred meters high, it would be possible to make an extremely good sub-20 Hz subwoofer. Finding something to play that would make use of it (other than early Metallica hits) would be hard, though.
The problem can be re-expressed as follows: The signal has very little kinetic energy or momentum. The laws of physics require both to be conserved. Kinetic energy is equal to the mass times the square of the velocity, whereas momentum is equal to the mass times the velocity. Anything not transferred is lost (as heat, sound, etc). If you know the mass of the air impacting the microphone and the mass of the moving components within the microphone, you can figure out the efficiency of the energy exchange.
You would have to build a device which was capable of oscillating at the required frequency which meets three criteria:
First, that as much energy as possible is transferred.
Second, that the energy required to change the state of the microphone (either from a stationary state, or from a moving state) does not exceed the energy that is transferred.
Third, that the system is only getting the energy you want and NOT picking up extraneous noise from more powerful sources elsewhere.
For the non-scientist, this loosely translates into: You've got to be able to hear what you want and because it's so faint in comparison, not hear anything you don't want to hear.
All this is simple enough in theory, but in practice it is actually very hard.
The difference between ADSL (one-way broadband) and SDSL (two-way broadband) is often just a setting in each of the routers. The software isn't any different, the wires aren't any different and I sincerely doubt DSL companies invest in multiple network interface devices to cover the full range of customers, when buying one in bulk will likely be cheaper and easier to reprogram.
The whole ADSL thing, these days, has grown tired. There is more than enough capacity to switch everyone to a full SDSL connection - and with the not-so-veiled threats from telecos, it might even be in the best interest of ISPs to push Internet technology to the absolute limits, even if it does cost them some profit in the short term. In the longer term, it might save their necks, which (I would have thought) they would consider rather more valuable.
I'd say #3 also applies, if the rootkit is hiding itself (and files) by modifying the OS' kernel or kernel configuration. Since that is generally (but not universally) how rootkits work, that option might also have some credibility.
(Especially to a non-technical audience, too, as Sony would then be put in the position of having to explain how their DRM did all this without modifying anything. There's no fifth to claim in the UK, and the right to silence was abolished. Hey, I detest those changes, but they cut both ways and I've no scruples about putting Sony through the wringer.)
...could probably be used in this way, for this software. The program was unquestionably not authorized by the user, as it is not declared in the EULA. As there is no apparent (yet) "Phone Home" capability, it would not violate the Data Protection Act. It might violate tresspass/break-and-entry laws, as the only reason the hacker of Prince Philip's e-mail account escaped conviction was that a transient tool was not considered a lockpick. This is a permanent tool that permits repeated intrusion, so I would guess the courts would be more sympathetic to the argument that it was breaking and entering. IANAL, but most people in computing in Britain have covered the DPA and CMA to some degree, because these are things IT people need to be careful of. It is possible - though unlikely - that the EU could also prosecute Sony over this, as it may infringe on privacy and computer protection laws in Europe. It's very doubtful the EU would take such action - they barely took any action against Microsoft for anything it did - but if Sony or other companies agravate the situation enough, there ARE elections in Italy coming up and the ruling elite there could do with someone to victimize.
America - well, there's no privacy in the US of A. The trade in personal information is open and widespread. There is an excellent chance that if anyone tried to prosecute Sony over privacy infringements that it would be laughed out of court. You can't protect what you don't have. Posession is 9/10ths of the law, and Americans posess very little - much as they often like to believe otherwise.
Sony actually has a much stronger case. Reverse-engineering their DRM scheme is in direct violation of both the letter AND the spirit of the DMCA, which is explicitly intended to prohibit exactly this kind of research (ie: the study of the spyware) and this kind of result (ie: the removal of it, afterwards). Depending on who Sony licensed the rootkit from, there is a possibility it might also violate aspects of the PATRIOT act. (If the rootkit is also used by any law enforcement groups, then this study could compromise wiretapping provisions in the act.)
I would be in favour of a classification system that DID mean something. For example, we know that the composition of all the rocky planets we've investigated and all of the Gas Giants is mixed. We also know they have a single core. Neither of these is proven (or disproven) for Pluto, but could be tested. We know that the composition statement is false for all asteroids and the single core is false for all comets. It would seem easy enough to base a classification system on such parameters.
Would this help us any? Probably, yes. Because planets are of mixed composition, they must have formed in the very early accretion disc from the sun. Because asteroids and comets are relatively uniform, they must have formed AFTER centrifugal forces had separated out the elements - lighter elements to the outside (which is why comets contain a lot of hydrogen) and heavier elements towards the center (asteroids are based on iron and nickel, depending on location).
The label, by this scheme, would then indicate composition, structure and time of formation, as these three properties are inter-related. On the other hand, we can go by mass or diameter and learn relatively little - which I suspect is the way the IAU will go, because that's something astronomers can measure easily. Easy != (interesing || useful). In this case, easy is pretty useless and will be subject to future argument.
I'm sure there are better methods of classifying, but I firmly believe the only useful method of classification is one that will allow predictions to be made and tested. The periodic table of the elements, for example, as a way of depicting valence theory is exceptionally useful. You can make useful predictions about groups of elements or even individual elements, based on the position in the table. Astronomical classifications should be no less useful and (given that we've far more powerful ways of obtaining, classifying and representing data today than early chemists) really should be a far MORE (Moore?) powerful tool.
Slashdot Mirror
It would have been sooo much better if they'd said "ten long horns"
I am now convinced that someone at SCO has flipped their lid and become a paranoid schizophrenic. Either that, or they are aiming at a career on the Comedy Channel once SCO sinks without trace.
They get it that they can scam, bilk and price-gouge legally, so long as it is in small amounts at a time.
They get it that they're going to be able to milk the compliant viewer for all they're worth (and then some), boosting profits and keeping the shareholders happy, while keeping the programs sub-standard and their integrity non-existant.
From the perspective of the TV companys, what is there not to get?
TV stations were being relayed over CU-SeeMe and the Multibone over ten years ago. Legally. Without advertising. Free. This "new stuff" is old-hat - or would have been if users had pushed for the technology to be widespread. As it was, CU-SeeMe denegrated into porn and commercialism, then vanished into the abyss. The multibone persists, but virtually no ISP is being encouraged to carry it.
If the site is very small and the database code is not particularly advanced, you might even do better with SQLite. However, this does require using PHP 5. If you're on PHP 4 or earlier, that's not an option.
Now, this changes dramatically if you need very advanced database features. Postgres is certainly far more advanced than MySQL (although the gap is not as great as it once was). It can also scale better, but for a small site that probably won't matter so much. A website that allows you to manipulate geographical data could be small/hobbyist but would still require an awesome database to drive it.
It all depends on what you're doing.
Databases are often used for dead projects, buried under large mounds of data. The parallel would seem to be pretty exact.
(No, this company isn't likely to hold onto it. This looks like the sort of deal where they buy something that might be worth a lot more next week or next year, so they can sell it and make money. Simple speculation.)
Note that this theory ONLY holds up if the price was sufficiently low that if Open Source does have a boom, the investors will be able to make significantly more profit on it than they would have done speculating on, say, stocks or currency. So if it's a land-grab, they're expecting the value of Open Source software to at least double or triple in the next few years. Otherwise, the investment wouldn't be worth it.
(The value of Ingres is important in this - if it cost them $10, then a 100% profit margin won't be hard to achieve. The more they actually paid - never mind anything officially said - the more it will take to make the profit margins they'd have to be looking at.)
There is one other possibility, but it's a remote one. Instead of a boom, they're expecting a crash - particularly in the database market. They'll have IT guys and the Open Source mantra of being able to look at the code can be persuasive. They may be convinced that the database market is so glutted and so unsure of direction that it will have to collapse. If it does so, then the only support they'll have is whatever they give themselves anyway.
In this scenario, they're playing the role of survivalists - acquiring the technology they'll need to survive, expecting things to get nasty for everyone else (ie: everyone still on their feet playing the role of SCO). In this scenario, the only sure defence is to have something with no prior technology in it - and Ingres would meet that. This does seem unlikely, but I wouldn't rule it out without a lot more information.
If you want something that'll do a bit more, but retain a lot of the speed and also keep the footprint down, MySQL is probably the best bet. It has a lot of the functionality of the really large databases - perhaps rather more than is good for a lean, mean database machine - but still gives good balance between function and performance.
PostgreSQL is much more powerful than MySQL, but at the cost of being bigger and generally not as nimble. If you're dealing with mid-sized Enterprise-level databases, I frankly wouldn't care about per-transaction performance as much as I'd care about maintaining good performance for respectable databases for a few hundred simultaneous users. MySQL would be hammered long before PostgreSQL for this kind of work. For GIS stuff, where you've some fairly tricky topographical information to manipulate, you'll notice that PostgreSQL has a far bigger following.
This leaves Ingres, which has a reputation for being good for Big Corporate Data Warehousing. Data warehousing is a very different problem from regular relational database handling. The problem-space is typically orders of magnitude greater, the database generally isn't going to be normalized and the mappings are generally altered to be less I/O-bound - which usually means more work for the CPU.
The problem-spaces solved by these databases are all very very different. I would love to see a database that had pluggable components such that different components were optimized for different types of workload and that different functions could be loaded/unloaded as needed, so your footprint was always the lowest for what you were doing, not what the database was capable of.
The fact is, no such database is in wide use - assuming it exists at all.
And we're only talking about the SQL Relational model. The "pure" relational model (as discussed on Slashdot many times) is different again, as are the Object-Relational and the Object-Oriented models. Absolutely none of these could be used with Xanadu's ZigZag data structures (the relationships are essentially permutations and order-independent, whereas formal databases use relationships that are either one-way or follow a specific ordering). They generally don't distribute well, either, as data is bulkier than code, forcing you to load-balance rather than distribute logically.
(That last part can be solved for some cases - it is fairly common to have a mix of tightly-coupled and loosely-coupled data, so it is possible to split the problem-space in a way that keeps communications down and takes advantage of a parallel architecture. It just isn't easy.)
Ingres is important as a database, because PostgreSQL can't (yet) cut it in the Really Big Database World. As good as PostgreSQL is, I would not want to replace extremely large-scale Informix or DB/2 databases with it. Maybe someday, but not today. Ingres - I'd give it a maybe. It does have the reputation of being able to handle it.
The multitude of engines out there are also important, because there are many, many different problem-spaces out there and NONE of them - not one - is good at even a few, let alone many, and certainly not all. And there are many problem-spaces for which there are no databases at all. At least, none worth mentioning.
This is a fixable problem, but not until someone goes out there and fixes it. That isn't happening. So, until then, I'll use RRD, SQLite, MySQL, PostgreSQL, Ingres, Sleepycat DB4, ZopeDB, OpenLDAP, CUPS*, Shadow*, Postfix*, Reiser4*,
*These all have databases in them. The password file, the print spooler, all the fancy file-access features of Reiser4... And because they're all working in subtly different ways, you WILL have a database engine for each, until or unless someone produces a system that can do all of these as well as each of the specialized solutions.
This should terrify celebrities the world over - if something this small and elusive can't escape the mercenary photographers of the tabloid press, what can?
Does this mean we should take every crackpot seriously? No, of course not. What it does mean, though, is that serious scientists should be encouraged to not rule something as meaningless merely because it is likely incorrect as stated. If we'd done that, we'd never have learned anything about anything.
A trivial example is the "Cold Fusion" fiasco from Fleich and Pons (spelling may vary) from Utah. Well, I doubt anyone seriously expected anyone from Utah - especially chemists - to stumble onto anything interesting in physics. And, surprise surprise, they didn't. What they DID stumble onto, however, was a very interesting form of fuel cell that can store fairly large amounts of hydrogen within the cell.
ObTrivia: The problem with Apollo 13 was that hydrogen and oxygen stored for use by the fuel cells was vented into space after an explosion. Conclusion: If the fuel was stored in a chemically stable form, which could be electrically released to generate more power than was used to release the fuel, then you'd have an fairly accident-proof fuel cell. If the fuel was then contained wholly in the cells, you would need no fuel tank or fuel lines, removing a problem with existing hydrogen technology.
Can these claims have any meaningful value? I don't know, but I do know that if they do and they are 100% ignored because they're meaningless as is, we never will know. The trick is to learn what is useful without being burned by the useless, discarding that which cannot be usefully learned from without discarding information which would save time to examine closer.
Now, I can see key lengths increasing - hashes are up to 512 bits, so 512-bit keys would seem a logical step. NIST are researching encryption modes that provide a much higher level of security, and this is another area I see getting a lot of attention in the future.
So encryption is definitely not a dead subject and I think AES will be seen as naively weak in the sort of timescale you're giving (a century or so). Even Serpent (another AES contender) is only rated as secure for another 50 years unless the algorithm has been broken before then.
Multi-pass encryption with multi-pass modes that are tamper-resistant and repudiation-resistant would seem the next logical step in encryption technology. Multi-pass is good, because encryption algorithms don't randomize sufficiently and it is often possible to extract some contextual information.
I also think it likely we'll move away from symmetric ciphers to asymmetric, provided a good parallel algorithm can be found. CPU cycles are cheap, these days, so the old excuse that public key encryption was slow is no longer so valid. A solid parallel algorithm would demolish that reason altogether.
First, the British police probably can't use the NSA to do decryption work. Second, unless the NSA has actually broken the algorithm, they are simply not going to be capable of breaking any code longer than about 64 bits in the useful lifetime of the data. If the encryption algorithm is a one-time pad (or a close faccimilie thereof), then they will be incapable of breaking it at all without the key. OTPs are, quite literally, unbreakable because EVERY possible decryption is equally likely.
Such a computer can break an ordinary (56-bit) DES key in 18 hours, 12 minutes and 16 seconds at worst. The average time to break a DES key on such a machine would be 9 hours, 6 minutes and 8 seconds.
To break a 128-bit key would require the computer to run for 2^88 seconds, or 9,813,705,283,528,192,184 years.
If you're willing to relax the rules a little - require no obvious precursors - then the number of innovations grows considerably. But they're still pretty thin on the ground. I'd say that there would be maybe one or two such innovations every decade, somewhere on the planet.
If you go further in relaxing the definition, and merely require no trivial precursors (ie: it cannot be a simple extrapolation or interpolation of what already exists), then there are maybe half a dozen each year.
I tend to go somewhere between those last two cases, in the way I look at innovation. I want the term to be used often enough to be meaningful to people, without being so broad as to be meaningless within itself.
However, I would like a clear term that is expressly for the rare cases that meet the first definition. A spectrum isn't just defined in terms of the progression along it, it is also defined in terms of the extreme ends. Calling them "totally unoriginal" and "totally original" tells you nothing other than they're points on the originality continuum. Zero and infinity, on the number line, are not just points. The expression of zero unleashed whole areas of mathematics, as did the expression of infinity. Their value in defining the limits vastly exceeded the value of any other point on the number line.
What this argument over innovation does is tell us about the line of creativity. It does not define the limits and we're all still arguing over the relative merits over some small segment of the line in its entirity. I think it's time to study the line, the dynamics of it, the maths of it, and the limits of it. Do that and we won't just agree on what is innovative, we'll be able to express the innovation in a way that has meaning.
It's still a derivative, in the sense that it is not wholly original, but it is not wholly derivative, it has innovative aspects that were ingenius and original. To me, it is one of the clearest cases of where you can have it both ways.
I would argue, however, that later generations can be innovative - provided they do something revolutionary in and of itself. For example, CERN's webserver was the first (and therefore innovative by being first) but I'd consider NCSA's webserver to to have posessed qualities that the CERN server did not have, in a manner such that NCSA's webserver (IMHO) deserves the title of innovative as well. Although Apache has yet more qualities, I would not consider those to be in a manner that justifies such a title.
By implication, I'm saying that a quality must have some attribute that distinguishes itself above and beyond being a mere addition, for the idea/project to be called a true innovation. To me, that attribute is that the addition not simply be an extrapolation or an interpolation of what already existed but must exist outside of the covered space, yet intersect the covered space in such a manner that the extension is a natural extension, not forced.
The "dumb person's test" for true innovation is that it must be so difficult to see in advance that it had truly occured to nobody at all. EVER. Yet be so obvious once found/developed that nobody really realizes it hasn't always been there,
There are a number of interesting research OS' out there, but I couldn't tell you if any of them were derivative or not. uMicro's Object Oriented OS in the late 80s was pretty interesting. Slow, limted but interesting.
The Transputer/Occam coupling was intriguing in that it provided a layer that wasn't strictly an OS, but provided the system calls an OS would have. As such, it could be considered totally non-derivative, as there was nothing there to be derived. I miss the Transputer - it was by far the best design of processor for its day, and its fate was a miserable, undeserved one.
What you need to do is move a far, far larger column of air. There has been work on using electrical sparks to generate pulses of sound which you can then modulate. This is not mechanical, so you get better efficiency on converting to sound. It also has far better response, would be capable of handling larger amounts of energy, and the weather systems have been demonstrating a very large-scale subwoofer for billions of years.
In all seriousness, I do believe that if you built your own "lightning generator", with diaphram, on the order of perhaps a few hundred meters high, it would be possible to make an extremely good sub-20 Hz subwoofer. Finding something to play that would make use of it (other than early Metallica hits) would be hard, though.
You would have to build a device which was capable of oscillating at the required frequency which meets three criteria:
For the non-scientist, this loosely translates into: You've got to be able to hear what you want and because it's so faint in comparison, not hear anything you don't want to hear.
All this is simple enough in theory, but in practice it is actually very hard.
The whole ADSL thing, these days, has grown tired. There is more than enough capacity to switch everyone to a full SDSL connection - and with the not-so-veiled threats from telecos, it might even be in the best interest of ISPs to push Internet technology to the absolute limits, even if it does cost them some profit in the short term. In the longer term, it might save their necks, which (I would have thought) they would consider rather more valuable.
...the temperature being in the 60s is that the processors can't handle any more LSD.
Now, some of those are "well-known", but there are plenty there that few people (even on Slashdot) are likely to be overly familiar with.
(Especially to a non-technical audience, too, as Sony would then be put in the position of having to explain how their DRM did all this without modifying anything. There's no fifth to claim in the UK, and the right to silence was abolished. Hey, I detest those changes, but they cut both ways and I've no scruples about putting Sony through the wringer.)
America - well, there's no privacy in the US of A. The trade in personal information is open and widespread. There is an excellent chance that if anyone tried to prosecute Sony over privacy infringements that it would be laughed out of court. You can't protect what you don't have. Posession is 9/10ths of the law, and Americans posess very little - much as they often like to believe otherwise.
Sony actually has a much stronger case. Reverse-engineering their DRM scheme is in direct violation of both the letter AND the spirit of the DMCA, which is explicitly intended to prohibit exactly this kind of research (ie: the study of the spyware) and this kind of result (ie: the removal of it, afterwards). Depending on who Sony licensed the rootkit from, there is a possibility it might also violate aspects of the PATRIOT act. (If the rootkit is also used by any law enforcement groups, then this study could compromise wiretapping provisions in the act.)
Would this help us any? Probably, yes. Because planets are of mixed composition, they must have formed in the very early accretion disc from the sun. Because asteroids and comets are relatively uniform, they must have formed AFTER centrifugal forces had separated out the elements - lighter elements to the outside (which is why comets contain a lot of hydrogen) and heavier elements towards the center (asteroids are based on iron and nickel, depending on location).
The label, by this scheme, would then indicate composition, structure and time of formation, as these three properties are inter-related. On the other hand, we can go by mass or diameter and learn relatively little - which I suspect is the way the IAU will go, because that's something astronomers can measure easily. Easy != (interesing || useful). In this case, easy is pretty useless and will be subject to future argument.
I'm sure there are better methods of classifying, but I firmly believe the only useful method of classification is one that will allow predictions to be made and tested. The periodic table of the elements, for example, as a way of depicting valence theory is exceptionally useful. You can make useful predictions about groups of elements or even individual elements, based on the position in the table. Astronomical classifications should be no less useful and (given that we've far more powerful ways of obtaining, classifying and representing data today than early chemists) really should be a far MORE (Moore?) powerful tool.