Re:I'm worried about this. It could be a disaster.
on
Microchips That Evolve
·
· Score: 1
Genetic programming
has been going on quite a while. I don't understand why people get so
excited when an old idea gets implemented in hardware. Doing
something in hardware instead of software is just an optimization. It
runs faster, but
it still does pretty much the same thing.
The system you want does exist. It's called the Open Directory Project. You can access it via a defined network protocol (HTTP) and it has an easy way to add and change entries (web forms). I think it just allows name collisions at the moment, but really that's fine since the kind of names you're proposing (descriptions of real world names) are not necessarily unique.
What does this have to do with computer science? The complaints mentioned in the article were were really about how the software industry was screwing up. I don't think it's fair to blame computer scientists, since CS research is generally ignored by software developers.
It might be fair to blame the industry for not making usability a priority, but it's generally a low priority for customers too, and companies prefer not to spend resources on features customers don't care about.
Xmcd has always used track lengths and offsets to calculate the id. The fuzzy matching may have changed, but has existed some form for years.
Here's an excerpt from the Xmcd changelog:
Ver 1.3 released Fri Sep 30 12:32:10 PDT 1994
The xmcd database Link feature CD list now displays the average track offset difference between the current CD and each potential database match. Those entries that are within 10 seconds are now listed with bold font. This enhancement is based on code contributed by Andreas Zins [gives email address].
Regardless of the GPL, the older versions of Xmcd exist as prior art. You only get one year after publication (which includes code releases) to patent something.
We could create a forum (web page) where independent musicians could claim to be simillar to famous artists, provide mp3s of their music, and lisseners would vote on the simillarity.
Or we could do it automatically with DSP and AI magic.
How many defects will they typically ship with? Many "acceptable" LCDs (acceptable to the manufacturer, not necessarily the customer) have one or two pixels that are always on or always off. How many bad pixels will consumers of this screen have to live with?
I wish someone would write a client that speaks both
Freenet and Mojo Nation protocols...
I'm sure someone will. It's a pretty obvious and useful thing to do.
Another neat thing to do would be to make Freenet search systems
available on Mojo Nation. The search broker would accumulate a
reputation from its response time and people's ratings, and people's
block clients could use that (plus anything else) as a measure of
trust, and as also a measure of how much Mojo they'd pay to use it.
I don't think Mojo Nation is really set up to deal with things like
that, but it might be easy to hack it into a content tracker and a
quartet of a block servers. (Tracker gets request for
"search?Metallica", generates a response, hands the response to the
quartet of local block servers, and tells the client where to get the
blocks.)
There's been lots of other work done on this. I've put up some links on my own site, but rather than get swamped I'll copy them here. I'm doing my thesis on automatic music classification. I've been planning to start a free software project from it; I was going to wait until I finished my thesis (a couple months from now), but since we're all talking about it now, I went ahead and created a SourceForge project (project name "vole").
Is it really necessary to reduce music to a series of mathematic calculations?
It is if you want a software agent to help you organize it. It doesn't need to be a perfect summary of all the hopes and dreams reflected and evoked by the music. It just needs to keep your mp3 player from putting Carmina Burana right after those piano sonatas.
There are a lot of ways to organize music, and categories are just the simplest. The experiment used categories because it's easy to do and it's easy to measure how well it's working. The same measurements can be used for discovering music categories we never thought of, or for finding things that are similar in various respects. You can look for types of similarity that maximize classification accuracy, or by example (user provides examples of things they think are similar, you generalize from that), or even use arbitrary similarity measures that may not relate well to what people consider to be salient.
You're making a strawman argument, and unfortunately lots of people have responded to it rather than pointing out the real issue.
Touretzky claimed that code was a medium of expression, not that things expressed in it would necessarily be emotive. The point isn't that a few people get teary-eyed when they see consistent indentation. The point is that code is an excellent way of expressing Djikstra's algorithm, for example. Would you rather describe quicksort in code or in prose? How would you feel about legislation that prevented you from describing it formally?
That would be a decent description if there were only one hill, but that's generally not the case. (If it were, you wouldn't need a GA.) The two solutions that breed are often climbing different mountains, and the children end up somewhere between, which could be on yet another mountain, or it could just be crap. There are schemes to avoid this (speciation), but they're not often used because they cause other problems (effective reduction of diversity, more tendency to get stuck in local maxima (sticking with your up-is-good)).
Of course there won't be any aliens using cell phones in that frequency range. The ACC will have allocated the whole band to their radio astronomers. We need an astronomical null modem to swap the TX/RX freqs or we'll all just end up listening for each other.
Claiming that quantum cryptography will bring impenetrable privacy to the masses is rather naive. Decent encryption is already available. How many people use it? The only thing that brings security to the masses is ease of use. If it takes any effort at all, no one will use it. Besides, encryption is only a small part of security.
The problem is that ther are no free trial versions of those compielrs.
Not true. gcj is GPL'd. That's free enough. It was probably just a matter of time, but it would have been more illuminating to compare VM-java vs static-java than VM-java vs static-C++.
It would have been nice if he'd also tested Java compiled to native code (such as with gcj) . He says the the point of his tests is to measure dynamic vs static compilation, so his experiments would be better if that were the only variable. It would also help squash the myth that coding in Java requires using a VM at runtime.
Animals (or plants for that matter) that kill do so purely functionally. Even cats play with mice to teach their young hunting skills.
It's not that simple. I've had cats who killed mice and left them on my doorstep whenever I came back from a vacation. Individual cats play with mice because they enjoy it. Cats have evolved to enjoy playing with mice because doing so is useful. Behaviors exist because they were more functional than disfunctional in their original context. That's all.
It is humans that humiliate and torture each other and seek to inflict emotional pain on victims and their families.
The meme that we're the only ones to kill for sport is fun to say, and appeals to us enough that it's as widespread as the ideas that lemmings drown themselves and Craig Shergold wants you to send him a get-well card. Brutalization is useful in certain contexts for control and status. I doubt we're the only ones doing it.
Even if some people employ it when it's not specifically helpful, that hardly sets us apart. Press the lever and get a reward, not because pressing a lever is good now, but because pressing it under similar conditions in the evolutionary environment helped more than it hurt. Animals will get fat if you give them access to their favorite foods.
What sets us apart is that we're better at learning and communicating than other animals. That makes better able to build on the work of prior generations and so eventually dominate our environment, and it also means we're better equipped to supplement or override our hardwired tendencies, for better or worse.
Maybe Europe's money would be better spent improving science education. Inventing new devices is great, but we'd also benefit from better dissemination of what we already know, and in the long run, that might lead to more technology than a direct investment.
I wonder if the real gold mine will be the competitive advantage of providing better service through personalization. Some sites actually provide a service, rather than just advertising one. Some of those sites could actually work better if they adapted to the user. To some extent, all sites provide the service of information and could be enhanced somehow. Even simple things could be useful.
But why would you want to carry around a few dozen CDs? If you want it stuck to your car, get an empeg. If you want a portable, the Nomad Jukebox is probably the best thing out there. Too bad it doesn't support Linux.
With the general lack of Linux USB support, it's still looking like I'll have to build my own player if I want something decent.
Why would you want it based on Latin and Greek? How many people speak Latin? Lojban cognates are based on a mix of the most popular languages in use today, in order to make them more familiar to the most people.
As for online dictionaries, go for it. Apologies to the French language gestapo, but language is defined by usage, not by committee. Set up whatever you like. Popularize words as you will. If enough people like them and use them, they will be words. No one needs to approve your process.
Slashdot Mirror
Genetic programming has been going on quite a while. I don't understand why people get so excited when an old idea gets implemented in hardware. Doing something in hardware instead of software is just an optimization. It runs faster, but it still does pretty much the same thing.
The system you want does exist. It's called the Open Directory Project. You can access it via a defined network protocol (HTTP) and it has an easy way to add and change entries (web forms). I think it just allows name collisions at the moment, but really that's fine since the kind of names you're proposing (descriptions of real world names) are not necessarily unique.
What does this have to do with computer science? The complaints mentioned in the article were were really about how the software industry was screwing up. I don't think it's fair to blame computer scientists, since CS research is generally ignored by software developers.
It might be fair to blame the industry for not making usability a priority, but it's generally a low priority for customers too, and companies prefer not to spend resources on features customers don't care about.
How many defects will they typically ship with? Many "acceptable" LCDs (acceptable to the manufacturer, not necessarily the customer) have one or two pixels that are always on or always off. How many bad pixels will consumers of this screen have to live with?
I'm sure someone will. It's a pretty obvious and useful thing to do. Another neat thing to do would be to make Freenet search systems available on Mojo Nation. The search broker would accumulate a reputation from its response time and people's ratings, and people's block clients could use that (plus anything else) as a measure of trust, and as also a measure of how much Mojo they'd pay to use it.
I don't think Mojo Nation is really set up to deal with things like that, but it might be easy to hack it into a content tracker and a quartet of a block servers. (Tracker gets request for "search?Metallica", generates a response, hands the response to the quartet of local block servers, and tells the client where to get the blocks.)
Yes, but so what? Even good search engines are bad, but they're still much better than nothing.
Those would indeed be very useful, but they're Hard to measure.
It is if you want a software agent to help you organize it. It doesn't need to be a perfect summary of all the hopes and dreams reflected and evoked by the music. It just needs to keep your mp3 player from putting Carmina Burana right after those piano sonatas.
There are a lot of ways to organize music, and categories are just the simplest. The experiment used categories because it's easy to do and it's easy to measure how well it's working. The same measurements can be used for discovering music categories we never thought of, or for finding things that are similar in various respects. You can look for types of similarity that maximize classification accuracy, or by example (user provides examples of things they think are similar, you generalize from that), or even use arbitrary similarity measures that may not relate well to what people consider to be salient.
You're making a strawman argument, and unfortunately lots of people have responded to it rather than pointing out the real issue.
Touretzky claimed that code was a medium of expression, not that things expressed in it would necessarily be emotive. The point isn't that a few people get teary-eyed when they see consistent indentation. The point is that code is an excellent way of expressing Djikstra's algorithm, for example. Would you rather describe quicksort in code or in prose? How would you feel about legislation that prevented you from describing it formally?
That would be a decent description if there were only one hill, but that's generally not the case. (If it were, you wouldn't need a GA.) The two solutions that breed are often climbing different mountains, and the children end up somewhere between, which could be on yet another mountain, or it could just be crap. There are schemes to avoid this (speciation), but they're not often used because they cause other problems (effective reduction of diversity, more tendency to get stuck in local maxima (sticking with your up-is-good)).
No, see, it's wireless. You'd just have a giant repeater halfway between the civilizations.
Of course there won't be any aliens using cell phones in that frequency range. The ACC will have allocated the whole band to their radio astronomers. We need an astronomical null modem to swap the TX/RX freqs or we'll all just end up listening for each other.
Claiming that quantum cryptography will bring impenetrable privacy to the masses is rather naive. Decent encryption is already available. How many people use it? The only thing that brings security to the masses is ease of use. If it takes any effort at all, no one will use it. Besides, encryption is only a small part of security.
Not true. gcj is GPL'd. That's free enough. It was probably just a matter of time, but it would have been more illuminating to compare VM-java vs static-java than VM-java vs static-C++.
It would have been nice if he'd also tested Java compiled to native code (such as with gcj) . He says the the point of his tests is to measure dynamic vs static compilation, so his experiments would be better if that were the only variable. It would also help squash the myth that coding in Java requires using a VM at runtime.
I'm sure there are plenty of radio astronomers who would love to see the Iridium constellation shut down for good.
It's not that simple. I've had cats who killed mice and left them on my doorstep whenever I came back from a vacation. Individual cats play with mice because they enjoy it. Cats have evolved to enjoy playing with mice because doing so is useful. Behaviors exist because they were more functional than disfunctional in their original context. That's all.
The meme that we're the only ones to kill for sport is fun to say, and appeals to us enough that it's as widespread as the ideas that lemmings drown themselves and Craig Shergold wants you to send him a get-well card. Brutalization is useful in certain contexts for control and status. I doubt we're the only ones doing it.
Even if some people employ it when it's not specifically helpful, that hardly sets us apart. Press the lever and get a reward, not because pressing a lever is good now, but because pressing it under similar conditions in the evolutionary environment helped more than it hurt. Animals will get fat if you give them access to their favorite foods.
What sets us apart is that we're better at learning and communicating than other animals. That makes better able to build on the work of prior generations and so eventually dominate our environment, and it also means we're better equipped to supplement or override our hardwired tendencies, for better or worse.
Maybe Europe's money would be better spent improving science education. Inventing new devices is great, but we'd also benefit from better dissemination of what we already know, and in the long run, that might lead to more technology than a direct investment.
I wonder if the real gold mine will be the competitive advantage of providing better service through personalization. Some sites actually provide a service, rather than just advertising one. Some of those sites could actually work better if they adapted to the user. To some extent, all sites provide the service of information and could be enhanced somehow. Even simple things could be useful.
But why would you want to carry around a few dozen CDs? If you want it stuck to your car, get an empeg. If you want a portable, the Nomad Jukebox is probably the best thing out there. Too bad it doesn't support Linux.
With the general lack of Linux USB support, it's still looking like I'll have to build my own player if I want something decent.
That looks nice. It'd be even nicer if they were actually shipping.
Why would you want it based on Latin and Greek? How many people speak Latin? Lojban cognates are based on a mix of the most popular languages in use today, in order to make them more familiar to the most people.
As for online dictionaries, go for it. Apologies to the French language gestapo, but language is defined by usage, not by committee. Set up whatever you like. Popularize words as you will. If enough people like them and use them, they will be words. No one needs to approve your process.