I don't see anything that makes this obviously impossible, but I'm still not convinced of its truth. Among other things, it's a story submitted by the inventor himself, which smells a bit funny to me. (Then again, if I invented something really cool and wanted to tell the world, I might well use Slashdot.) I think I'm also uneasy about the complete lack of information about how the method works. I suppose it's necessary for patent purposes, but I think it might have been a bit premature to submit "news" that doesn't actually inform us very much.
Anyway, I hope this is for real, but I'll believe it when I see more than an arbitrary signal trace.
It would be nice if there was some kind of `standard' letter that could be copy/paste/emailed to the patent office. A lot of people stating the same always attracts attention and it will show that there is some consensus in this community.
As someone who's worked for Congress, I can tell you that a form letter is exactly what you do not want to send. When I was sorting mail to the various staffers, there was a special bin for postcards. Letters would get someone to read your letter and tell me to send you a form response back. Postcards got you a postcard saying "Thanks for sending us that postcard,"; at the end of the day I'd count up the postcards for and against the vote.
A week or two after I started working there, I started seeing letters that all had the same text: they were instances where someone had taken a lobby group's sample text and copied it verbatim. These letters were not sent on to a human to read; they were sorted as postcards. That's far less influence on the mind of the decisionmaker.
Even though form letters would allow us to mount a massive campaign, it's my experience that you can get better results with a smaller number of impassioned individuals writing personal messages.
Genes are not patentable - applications based on a gene sequence are
False. Or rather, ought to be true, but false in practice. Consider, if you will, this press release announcing a patent on a key HIV cellular receptor. Nominally, the patent does cover "techniques" related to it, but the techniques listed on the front page are all just references to standard molecular experiments that can be performed when you know a gene. The basic argument seems to be that the gene is a process for producing a protein, and is therefore allowed. (View front page of patent at IBM.)
It's begun. I think the patent system is great, but genes and software (the two new and rapidly-growing fields) are not handled properly. Too many people stand to gain from the current system for it to be fixed. IMHO, the correct thing to do would be to take a patented gene (the sequence must be disclosed in the patent, and all you need is that sequence, a lab, and a few grad students...) and come up with some amazing cure. Announce it to the world. State exactly how to use it. The patent-holder will sue. You then get to go before a court and explain that BigGeneCorp is holding back treatment that could save thousands. Eventually, you will go to the Supreme Court, which will probably throw away all such patents because they're secretly a bunch of libertarians. It'll cost millions and take a decade, but genes will be free again.
There's also a simpler but less powerful attack. This patent covers one sequence of nucleotides. Most human proteins can be encoded by more than one sequence. Take this sequence and find a spot that (according to computer modeling) allows polymorphisms. Express that new protein to be sure it works just as well. Now patent that and make the patent freely available. This can, in theory, be done for almost any patented gene. (This is a business model. You may use this business model which I have just invented if you do not attempt to profit from it. If you try to profit from it, I will patent your grandmother and have you killed for infringement.:-)
Alik
Re:Complete lack of privacy already!
on
Database Nation
·
· Score: 1
The other day I went to the dentist and had to fill out one of those interminable insurance forms. Amongst other things, they wanted to know if I'm single/married/divorced/separated etc. Why the HELL do they need to know that? When are they going to start enquiring about my sexual orientation, as well as my preferred sexual positions?
Actually, there is a reason for any MD to ask you those questions: epidemiology. A big thing these days is "evidence-based medicine" --- the idea that maybe doctors should base clinical judgements on good statistics rather than their own faulty memories and biases. (What? Medicine should be based on rational scientific technique? What a new idea!)
Married people have different habits than single. If some weird oral condition shows up, it'll be useful to see what variables are associated with it. This kind of weird clustering is how we discovered that fluoride prevents tooth decay --- dentists were seeing kids with mottled teeth in some towns, and also seeing way fewer cavities in those kids. Do some studies, find fluoride as the key factor, and suddenly most of the population still has a full set of teeth.
Is it intrusive? Yes. Should insurance companies (or doctors) be allowed to do whatever the hell they want with that? Fuck no. However, it can be useful. (I do agree that there should be a disclosure with each question saying why it's being asked. Perhaps hypertext will make this a reality once people grow some brains.)
Almost exactly one week later, I must retract my comment and call myself an idiot for believing that government has any sense. The US government really does allow people to say "We found the gene for X, so we have dibs on all commercial stuff involving X." Here's the press release I'm thinking of; I just submitted this as a story as well (as, no doubt, did three trillion others).
It's an interesting bit of performance art, but practically speaking, nobody needs to do this. No existing government would allow a company to say "We found the genes for lung development, so everyone pay us royalties." Sure, a government that's been completely taken over by the company in question would allow it, but if that's happened, how much protection is a patent going to give you?
What will happen is that companies will be allowed to say "We found the genes for lung development and have created a cure for cystic fibrosis. Pay for it." Honestly speaking, I think that's fair. There's a reasonably large investment of time and money that has to be made to sequence genes, work out what they do, and turn that into useful therapy.
Most gene patents to date (or at least the ones I personally have seen) are along the lines of "This is a sequence which allows you to perform this operation on the genome. Pay us and we'll sell you copies of it." What you'll get in the vial they send you is a molecule that one is unlikely to find in nature. To me, that means we can't complain that they're patenting a natural law.
I will admit that I do not trust the big biotech companies to be any nicer than Amazon or Microsoft or Sun. However, most of their current nastiness is in sneaking genetically modified foods around without bothering to tell anyone (at least in the USA). To date, I have not seen an abuse of gene patents. Until we know what form the abuse will take, we can't directly outlaw it. (As an analogy for those who've played RPGs, consider the attempts of a GM to keep players on a straight track. It never works. People will always find something different to do than you expected.)
Re:Is command line vs. GUI a false dichotomy?
on
Middle Media
·
· Score: 1
Once scripted, they essentially become powerful new commands. I haven't seen a GUI that privdes this mechanism to script repeated operations and build up complex combinations with conditional logic. Is there one?
MacOS comes to mind. Most applications can speak AppleScript, which allows you to create reasonably powerful system scripts. I personally find AppleScript's syntax to be annoying because it tries to be "English-like" and I don't think in English when I write programs. In general, I have not been able to leverage AS in the same way I have Perl. However, it does meet your criterion of a scriptable GUI.
No sources named. In a paper which has been pointed out by another poster to have the tabloid nature.
You know what this probably is? Someone in France heard about NSAKEY. This got mentioned in a report as "Probably not actually related to the NSA, but we're putting a note on it in the We-Hate-Microsoft file." This in turn gets transmitted via friend-of-friend through three or four hops, reaches The Age, and you've got this lovely article.
Does anybody else remember a year or two ago when telomerase was figured out? You couldn't throw a rock without hitting articles about how this was the key to the fountain of youth and various Katzian pontifications on what it will be like to be immortal. Funny, I don't seem to be immortal. Wonder what happened?
As far as I can tell from this story, they've found a transcription factor that exists in both yeasts and mammals. This is nice biology and will certainly be important, but there are many evolutionarily conserved transcription factors, none of which is the "secret to aging". Similarly, there are many reactions which happen to require NAD. Fat synthesis requires NAD, and if you take away NAD, the synthesis stops. This does not mean that the acetyltransferases and dehydrogenases are the "secret to aging".
In short, I don't see anything here that makes this protein particularly special. The title of the article is pure hype.
I really wonder what these are going to be like. Will it just be skins and hotlists? Things like "customized to work very well with CNN content" make it sound like companies could muck around with the code a bit and sell/distribute the derivative work with the Netscape name still on it.
This has the potential to be pretty cool. Joe Windows distrusts that long-hair hippie browser Mozilla 'cause it's written by hackers, but he knows Netscape and AOL are names to be trusted. He could probably be convinced to use "CERT Netscape: The most secure web browser available", or something like that. This could be a really good way to get secure privacy-protecting browsers into the hands of the Unwashed Masses.
yeah, I've actually heard that some people get trepanning for recreational reasons now - some sort of high if it's done over the right area of the brain. I don't know if it's true.
It's true. Consult www.trepan.com if you want to know why they do it.
Personally, I consider this to be ridiculous. They're basing their whole idea around one ancient paper by a Dutch medical student. Speaking as a medical student, I wouldn't trust any paper authored solely by anyone in my class, including myself. However, people are certainly free to put holes in their heads if they want.
Getting back to the topic of drugs and expanded reality: if they do nothing else, many "consciousness-expanding" drugs do temporarily alter your frame of reference. This may be a good thing, since it shocks people out of complacency, which will in turn lead to more active thinking. On the other hand, if you can't tell what's real and what's chemical, you're not very functional, so I doubt we're going to see a society where people do much hacking while intoxicated. I've seen code that I wrote while in an altered mental state, and it's utterly awful.
Your science is very flawed. For one thing, the 4 bases of DNA are not the only ones that can be used. Consider uracil, used in place of thymidine in RNA. Many things can attach to a deoxyribose-phosphate backbone. Our cells have a large population of "repair enzymes" which keep the system within the known limits.
Messenger RNA is not assembled in units of 3. An mRNA transcript may be of arbitrary length in theory. The actual transcript length is controlled by various other proteins besides RNA polymerase which bind to the DNA and direct initiation and termination of transcription. mRNA must also undergo extensive post-processing (more special proteins and sequences) in order to become something that can be theoretically translated to protein.
Many things besides RNA can bind to DNA. Most of these things are proteins your body uses for specific functions. Others are various toxins and damaging agents which must be removed by the repair pathway. Many things besides amino acids can bind to RNA; again, this is mostly protein, but can be other things. Amino acids do not spontaneously form large polypeptides. If they did, your cells would not have the nice stockpile of free AAs which they currently possess.
In short, the system is significantly more complex than you think. A college bio course does not get into even 1% of what's going on. My medical molecular genetics course maybe got into another 1% beyond that, and that's being charitable. The mechanisms of cellular replication (which are found even in the simplest bacteria, and to some degree even in viruses) are not a trivial system to build. Please do not offer grand opinions about molecular biology without really understanding how little even the great scientists know about it.
Personally, I find that the complexity of the system leads me to believe in the existence of God. It also leads me to have some ideas about the nature of God, but those are offtopic.
Cells may enter quantum states when they are unable to divide and replicate and become isolated....
Um, no, that's when they d-i-e.
Not at all true. Consider your neurons. They do not divide. (There may be some exceptions; the jury is still out.) However, they do not die as long as they continue to receive electrical and chemical stimuli. In fact, most of the cells in your body are in this "senescent" state. An adult has only a few localized populations of continually active dividing cells.
I personally have no idea whether or not this guy is right, but it's a neat theory. As others have pointed out, let him test it; if he's wrong, we can laugh at him, and if he's right we can give him Nobels in Biology and Physics at the same time.
I am particularly concerned about things like legal notifications being sent to you via e-mail. For very important documents, even postal delivery is not good enough. Some require a return receipt, and some require identity verification (not so much for privacy, but to verify that delivery was made) for delivery. What mechanisms do we have in place, or just have, that can do all this?
Well, for return-receipt, I would suggest something along the lines of the confirmation scheme currently used by many listservs. You sign the initial document via web. They consult your listed contact address with a central key registry and send both a confirmation and some arbitrary bits to that email address. You then sign the arbitrary bits and bounce them back via email. It is now presumable that the order was in fact placed by the person to whom the signing key belongs.
Does this have security flaws? Yes. For example, it remains vulnerable to a man-in-the-middle attack between central key registry and merchant. However, this is a framework created in one minute; a security professional can no doubt generate something a bit more secure.
It can be done using existing protocols and algorithms. I personally would like to see personal keys which are significantly bigger than the 128-bit junk used in the average browser; 2K might hold the line for a few years. (Yes, I know that bigger keys mean more encryption time. How many documents do you actually sign per day? Most people I know don't get past the single-digits.) There are issues in setting up infrastructure, and these must be resolved before you can get me to use such a system, but I think they could even be resolved correctly if people actually bother to think.
I'm a bit surprised that the big corporations haven't pursued the market for Indian-language computing more aggressively. Many of them have large numbers of bilingual programmers brought over on work visas. Given the sheer size of the potential market in India, I'd expect them to leverage this advantage.
One thing to note (and a possible reason for the lack of translation) is that there is no single "Indian language"; as with Chinese, there are several regional dialects which don't always bear much resemblance to each other. In practice, English ends up getting used a lot of the time as a universal standard. If you walk through Bombay, most of the billboards you see will be in English. The few times I've been there, most of the shopkeepers understood enough English to successfully complete a transaction.
Now, out in the villages, this is not true. On the other hand, I doubt the people in the villages care whether or not they have Internet access just yet. They need sanitation, electricity, and other basic health-improving technology. Once that's in place, then worry about whether or not they can be happy Amazon consumers.
The New Scientist article is baffling because it seems to have things backwards. Our livers (in the unhappy event anyone is foolish enough to consume methanol - not ethanol) in fact convert methanol into formaldehyde and formic acid, not the other way round. It is this formaldehyde and formic acid that account for the toxic effects - notably permanent blindness and destruction of the nervous system and cerebral cortex (not real fun). I know the article waves its hands and says, oh, the process is reversible. Maybe...
Not "maybe". You'll notice that the article also says that they drive the process using NADH. NADH, for those who don't know, is a critical metabolic compound in the pathway from glucose to ATP. (ATP is the primary power source for most cellular reactions that require energy.) The alcohol dehydrogenase reaction, as shown in the article, is:
CH2OH + NAD+ CH2O + NADH
When you have high levels of methanol in your body, the equilibrium is driven to the right. The NADH is promptly removed to the electron-transport system of the mitochondria (or to various synthetic pathways), thus keeping the reaction going in the forward direction. If there's nothing in your system which can remove NADH and you add excess NADH, the equilibrium is driven to the left, transforming formaldehyde to methanol. The other two enzymes in this process are also dehydrogenases, and thus probably also do NAD/NADH conversion, and thus the same rule applies. This is not handwaving, it is basic chemistry.
Now. It is true that there is a question of where the hell one is going to get the necessary NADH. It is also clear that they have no clue yet how this will be done, only some theories. However, the basic chemistry here is perfectly sound. Most metabolic reactions are, in fact, reversible; your body likes to keep its options open.
For those of us who can't make it, can a transcript of the discussion (with obvious trolls edited out) be posted to Slashdot? Bruce is clearly popular with at least part of the community, so I'd say this is at least as relevant as a regular interview article.
I think you're quite right about Brooks' current goals and motivations, and I'm equally annoyed by them. I guess it gets more people all fired up about AI/robots, which is nice, but it also tries to tell them that all the problems have simple solutions which don't involve nasty math, which is incorrect.
More fundamentally, a major problem in AI as a field is that people keep trying to make it to human-level AI in one step, rather than clawing their way up the evolutionary ladder with ant-level AI, bee-level AI, lizard-level AI, rodent-level AI, and only then attempting higher mammals.
Why should we do that, though? Logically, all important aspects of intelligent behavior are contained within humanity; therefore, it's easiest to stick to that model. Furthermore, if we work with the lower animals, there's a problem in that they have different sensory modalities, different motion needs, different connection pathways, and so on. (At least, there is if you're doing AI by trying to reverse-engineer the brain, which is my personal favorite method.)
Moreover, what would bee-level or rodent-level or lizard-level mean? I don't actually see a distinction between those levels --- if you have a general solution to the problems of motion planning, target location, and self-preservation, then you can build pretty much any animal you want. (For some animals you'd also need to have manipulation routines in your motion-planning.)
Motion planning et al. aren't really in the critical path for intelligent systems, unless you buy Brooks' thesis that intelligence only exists in embodied form. (I don't. If you made Helen Keller into a quadriplegic, she'd still be intelligent, she'd just go insane.) They are in the critical path for robots, which are generally lumped together with AI because we want AI so we can build bots with it, but the two domains can be separated. Of course, this thread was originally about robots, where such things are very necessary; strangely enough, many robot folks are working on copying the lower animals.
It's certainly not "one step from solution", but at the same time, I think humans are the right level to be working at for the "intelligence" part of things. We have to break the human into subsystems and parcel out the subsystems to different researchers (all of whom will say that their problem would be easy if the other guys would just get theirs solved), but I don't see that we need to solve rats first.
There's hope, though. Game AI developers are struggling to build characters that can survive in tough environments. They're getting better at it. Work in the game AI field is getting to be better than academic AI; the gamers have a real problem and a real market.
I wouldn't go so far as to say that game AI is better. I agree with the previous poster that the game folks haven't exactly made new breakthroughs. They mainly seem to take advantage of the fact that they've got abitrarily perfect sensing, motion, aiming, communication, coordination, and so on. Beyond that, game AI still seems to mostly be preprogrammed formations and decision trees. (On the other hand, many real soldiers work that way.) Even Deep Blue wasn't a particular breakthrough in technique --- they took the standard search tree chess algorithm, tuned it for Kasparov's playing style, and threw Moore's Law at the problem. Blue Gene will be something similar for the protein-folding problem. (Not that I care --- protein-folding has to be figured out somehow, and whatever works benefits everyone.)
I think a good goal for game AI is just-too-hard AI: a system which is always a little bit better than you, and thus keeps forcing you to improve on your own. Especially for things like chess, a system like that (with skill implemented as a limit on search tree depth and possibly multiple search algorithms) would make an excellent trainer. Sadly, I doubt it could teach me to play good Quake, because as far as I can see human Quake players don't follow patterns.
The email in question said something to the effect of "The AC's karma is currently 1972. Way not negative." You are welcome to doubt and ask Rob to rerun the calculation; perhaps he accidentally inverted a sign somewhere. I choose to believe he did it right.
can you reproduce a human baby exactly? would you want to? maybe if people just valued individuality and life this whole 'reproducibility' thing would not matter
That's one thing when speaking of a human life (and there are still cases when I would want to duplicate a baby), quite another when speaking of a scientific project. The Cog project is a scientific research endeavour. One of the criteria for good science is reproducibility of results. If you can't do it again, it's not a particularly big breakthrough.
Yes, humans might be that advanced as to come up with new ideas, by contemplating a problem long enough, but even for us (limited) random behaviour can bring unexpected rewards. I don't think the first people to discover burned (i.e. roasted) meat tastes pleasant or the primate noticing that using rocks as hammers opens things, did so by reasoning.
I agree, but I wouldn't define such things as random behavior. I would define them more as accidental discoveries. For example, cooking probably didn't come about because a human randomly decided to throw meat into a fire; it more likely occured by accidentally dropping meat into the fire, or by finding an animal which had somehow been caught in a fire. The randomness is on the part of the world, not of the agent.
Intentional random behavior in the hopes of discovering something new and interesting seems to be to be the basic principle behind genetic algorithms. GAs are neat, but personally I find the stuff they produce to be pretty ugly and non-robust. (Of course, so were some early robot programs.) GAs are another technology that I think gets a bit too much hype; I can see a bitstring evolving into "Hello, world!", but I can't see one evolving into a stable Mozilla.
Just because you don't know the emotion algorithm in humans doesn't mean it doesn't exist.
That's fair enough. However, I would say that if there is an algorithm, it is certainly not an obvious one.
Let us not confuse lack of comprehension with self-involved flattery. We are not that complex of beings.
Here I start to disagree. I believe that we are fairly complex beings, at least for any human definition of complexity. "Complex", in common usage, translates to "Damn hard to do and/or understand." I think it's perfectly reasonable to state that the human mind is damn hard to build and/or understand. What systems do you see that are so much more complex than the body? The only ones I can think of are those in which humans are components, such as social systems.
Everything we are is matter, and therefore we have a protocol in order to be a living being.
I disagree with this statement, or at least with what I believe you mean by this statement. If you are saying that mind can be solely reduced to the motion of particles within the space defined by your brain, I disagree. I believe that there is such a thing as a soul. I also believe that the soul can be detected and measured, but that we do not yet know how to do so. This means that the soul is, in some sense, matter (or energy, which is the same thing), and therefore we are in fact purely material. However, in general, when someone says "All we are is matter," they mean it as a denial of the soul.
You assume our emotions aren't predictable simply becuase we are not smart enough to be able to, yet. Our algorithm is obviously far more advanced then:
Actually, it's a bit more than that. Let's assume for a second that you're right and that our minds are nothing more than the known matter within our brains and bodies. You cannot accurately predict what the behavior of that matter will be. The actions of the brain depend on the motions of single ions, which in turn depend on the motions of charged particles, which are fundamentally unpredictable by the Heisenberg principle. This is a good thing --- it provides an empirical proof of something approximating free will.
But, you can keep believing what you want because we won't figure it out in your lifetime.
That's a pretty bold assertion with no backup, and I suspect it of being said just to incite and inflame. How can you predict what science will do? How can you predict how long I might live? Personally, I intend to see the year 2200, and hopefully many years beyond that.
As much as it should be adressed (to make robotics are "serious" since;-> IMHO it is part of the question, as "intelligent behaviour" has to contain some sort of randomness and unpredictability.
I'm not quite able to parse the first part of your sentence (especially because of an unclosed paren), but I can handle the second part, and I'm not sure I agree. I don't think intelligent behavior has to be random or unpredictable. Humans are not random. Humans are, to some degree, unpredictable. However, much of that comes down to how we're designed; another large factor is emotion.
Our first attempts at AI and semi-intelligent robots will probably not have emotion, just because we have even less clue about how emotion works than we do vision. They may make expressions (this is something the Kismet project at MIT is apparently doing well) for cuteness factor, but I doubt they'll have anything we call "real" emotion. I believe that their intelligence will be largely algorithmic in nature, especially the motion-planning parts of it. Therefore, their movements will be totally or almost-totally predictable, as will their responses to stimuli.
I can see an argument that some randomness will be built in because humans would be uncomfortable with something that always moved in the exact same way, but that does not imply that intelligence requires randomness.
BTW, for anyone interested there is the COG project at MIT, doing research into a humanoid robot, it is VERY impressive.
"Any sufficiently advanced technology is indistinguishable from a rigged demo." --Anon of Ibid (or somebody)
Seriously, though, while I'm not accusing Rodney Brooks & Co. of being frauds, I do think they're getting credit for a lot more than they've really done. Reactive control is wonderful for simple systems. In the MIT robot competitions, the second-place robot is generally some simple reactive design. (At Dartmouth, the first-place robot was, but the last Dartmouth robot competition had a bot construction/program time of about six weeks.)
However, the robot that wins out is generally not using a Cog-like architecture; it's usually one of the designs that uses a more artificial method of intelligence and navigation. The problem with reactive control is that it's easy to build and debug for simple systems, but as you add more functions and layers the connections and interactions become too much for even an MIT mind to handle. Brooks even says as much in one of his early papers on subsumptionism.
There's also Brooks' admission in his own FAQ that even the simple ant robots they've built are utterly unreproducible. IMHO, it's not that great if you can build something once but can't tell other people how to reproduce it.
This isn't meant to be a whole bunch of doomsaying about reactive design --- I think it's appropriate in some circumstances, and I've used it myself when I needed a simple barely-effective hack. It's just that sometimes it feels like the Cog Shop and their academic relations would like people to believe that they're the ultimate solution to all of robotics, and I don't think that's going to happen. It's too hard to use their methods in a general and structured manner to build systems that have anything resembling guaranteed behavior.
Slashdot Mirror
I don't see anything that makes this obviously impossible, but I'm still not convinced of its truth. Among other things, it's a story submitted by the inventor himself, which smells a bit funny to me. (Then again, if I invented something really cool and wanted to tell the world, I might well use Slashdot.) I think I'm also uneasy about the complete lack of information about how the method works. I suppose it's necessary for patent purposes, but I think it might have been a bit premature to submit "news" that doesn't actually inform us very much.
Anyway, I hope this is for real, but I'll believe it when I see more than an arbitrary signal trace.
It would be nice if there was some kind of `standard' letter that could be copy/paste/emailed to the patent office. A lot of people stating the same always attracts attention and it will show that there is some consensus in this community.
As someone who's worked for Congress, I can tell you that a form letter is exactly what you do not want to send. When I was sorting mail to the various staffers, there was a special bin for postcards. Letters would get someone to read your letter and tell me to send you a form response back. Postcards got you a postcard saying "Thanks for sending us that postcard,"; at the end of the day I'd count up the postcards for and against the vote.
A week or two after I started working there, I started seeing letters that all had the same text: they were instances where someone had taken a lobby group's sample text and copied it verbatim. These letters were not sent on to a human to read; they were sorted as postcards. That's far less influence on the mind of the decisionmaker.
Even though form letters would allow us to mount a massive campaign, it's my experience that you can get better results with a smaller number of impassioned individuals writing personal messages.
Alik
Genes are not patentable - applications based on a gene sequence are
:-)
False. Or rather, ought to be true, but false in practice. Consider, if you will, this press release announcing a patent on a key HIV cellular receptor. Nominally, the patent does cover "techniques" related to it, but the techniques listed on the front page are all just references to standard molecular experiments that can be performed when you know a gene. The basic argument seems to be that the gene is a process for producing a protein, and is therefore allowed. (View front page of patent at IBM.)
It's begun. I think the patent system is great, but genes and software (the two new and rapidly-growing fields) are not handled properly. Too many people stand to gain from the current system for it to be fixed. IMHO, the correct thing to do would be to take a patented gene (the sequence must be disclosed in the patent, and all you need is that sequence, a lab, and a few grad students...) and come up with some amazing cure. Announce it to the world. State exactly how to use it. The patent-holder will sue. You then get to go before a court and explain that BigGeneCorp is holding back treatment that could save thousands. Eventually, you will go to the Supreme Court, which will probably throw away all such patents because they're secretly a bunch of libertarians. It'll cost millions and take a decade, but genes will be free again.
There's also a simpler but less powerful attack. This patent covers one sequence of nucleotides. Most human proteins can be encoded by more than one sequence. Take this sequence and find a spot that (according to computer modeling) allows polymorphisms. Express that new protein to be sure it works just as well. Now patent that and make the patent freely available. This can, in theory, be done for almost any patented gene. (This is a business model. You may use this business model which I have just invented if you do not attempt to profit from it. If you try to profit from it, I will patent your grandmother and have you killed for infringement.
Alik
The other day I went to the dentist and had to fill out one of those interminable insurance forms. Amongst other things, they wanted to know if I'm single/married/divorced/separated etc. Why the HELL do they need to know that? When are they going to start enquiring about my sexual orientation, as well as my preferred sexual positions?
Actually, there is a reason for any MD to ask you those questions: epidemiology. A big thing these days is "evidence-based medicine" --- the idea that maybe doctors should base clinical judgements on good statistics rather than their own faulty memories and biases. (What? Medicine should be based on rational scientific technique? What a new idea!)
Married people have different habits than single. If some weird oral condition shows up, it'll be useful to see what variables are associated with it. This kind of weird clustering is how we discovered that fluoride prevents tooth decay --- dentists were seeing kids with mottled teeth in some towns, and also seeing way fewer cavities in those kids. Do some studies, find fluoride as the key factor, and suddenly most of the population still has a full set of teeth.
Is it intrusive? Yes. Should insurance companies (or doctors) be allowed to do whatever the hell they want with that? Fuck no. However, it can be useful. (I do agree that there should be a disclosure with each question saying why it's being asked. Perhaps hypertext will make this a reality once people grow some brains.)
Alik
Almost exactly one week later, I must retract my comment and call myself an idiot for believing that government has any sense. The US government really does allow people to say "We found the gene for X, so we have dibs on all commercial stuff involving X." Here's the press release I'm thinking of; I just submitted this as a story as well (as, no doubt, did three trillion others).
O brave new world, that hath such people in it.
What will happen is that companies will be allowed to say "We found the genes for lung development and have created a cure for cystic fibrosis. Pay for it." Honestly speaking, I think that's fair. There's a reasonably large investment of time and money that has to be made to sequence genes, work out what they do, and turn that into useful therapy.
Most gene patents to date (or at least the ones I personally have seen) are along the lines of "This is a sequence which allows you to perform this operation on the genome. Pay us and we'll sell you copies of it." What you'll get in the vial they send you is a molecule that one is unlikely to find in nature. To me, that means we can't complain that they're patenting a natural law.
I will admit that I do not trust the big biotech companies to be any nicer than Amazon or Microsoft or Sun. However, most of their current nastiness is in sneaking genetically modified foods around without bothering to tell anyone (at least in the USA). To date, I have not seen an abuse of gene patents. Until we know what form the abuse will take, we can't directly outlaw it. (As an analogy for those who've played RPGs, consider the attempts of a GM to keep players on a straight track. It never works. People will always find something different to do than you expected.)
MacOS comes to mind. Most applications can speak AppleScript, which allows you to create reasonably powerful system scripts. I personally find AppleScript's syntax to be annoying because it tries to be "English-like" and I don't think in English when I write programs. In general, I have not been able to leverage AS in the same way I have Perl. However, it does meet your criterion of a scriptable GUI.
No sources named. In a paper which has been pointed out by another poster to have the tabloid nature.
You know what this probably is? Someone in France heard about NSAKEY. This got mentioned in a report as "Probably not actually related to the NSA, but we're putting a note on it in the We-Hate-Microsoft file." This in turn gets transmitted via friend-of-friend through three or four hops, reaches The Age, and you've got this lovely article.
Nothing new here, folks.
Alik
Does anybody else remember a year or two ago when telomerase was figured out? You couldn't throw a rock without hitting articles about how this was the key to the fountain of youth and various Katzian pontifications on what it will be like to be immortal. Funny, I don't seem to be immortal. Wonder what happened?
As far as I can tell from this story, they've found a transcription factor that exists in both yeasts and mammals. This is nice biology and will certainly be important, but there are many evolutionarily conserved transcription factors, none of which is the "secret to aging". Similarly, there are many reactions which happen to require NAD. Fat synthesis requires NAD, and if you take away NAD, the synthesis stops. This does not mean that the acetyltransferases and dehydrogenases are the "secret to aging".
In short, I don't see anything here that makes this protein particularly special. The title of the article is pure hype.
Alik
I really wonder what these are going to be like. Will it just be skins and hotlists? Things like "customized to work very well with CNN content" make it sound like companies could muck around with the code a bit and sell/distribute the derivative work with the Netscape name still on it.
This has the potential to be pretty cool. Joe Windows distrusts that long-hair hippie browser Mozilla 'cause it's written by hackers, but he knows Netscape and AOL are names to be trusted. He could probably be convinced to use "CERT Netscape: The most secure web browser available", or something like that. This could be a really good way to get secure privacy-protecting browsers into the hands of the Unwashed Masses.
Alik
It's true. Consult www.trepan.com if you want to know why they do it.
Personally, I consider this to be ridiculous. They're basing their whole idea around one ancient paper by a Dutch medical student. Speaking as a medical student, I wouldn't trust any paper authored solely by anyone in my class, including myself. However, people are certainly free to put holes in their heads if they want.
Getting back to the topic of drugs and expanded reality: if they do nothing else, many "consciousness-expanding" drugs do temporarily alter your frame of reference. This may be a good thing, since it shocks people out of complacency, which will in turn lead to more active thinking. On the other hand, if you can't tell what's real and what's chemical, you're not very functional, so I doubt we're going to see a society where people do much hacking while intoxicated. I've seen code that I wrote while in an altered mental state, and it's utterly awful.
Alik
Messenger RNA is not assembled in units of 3. An mRNA transcript may be of arbitrary length in theory. The actual transcript length is controlled by various other proteins besides RNA polymerase which bind to the DNA and direct initiation and termination of transcription. mRNA must also undergo extensive post-processing (more special proteins and sequences) in order to become something that can be theoretically translated to protein.
Many things besides RNA can bind to DNA. Most of these things are proteins your body uses for specific functions. Others are various toxins and damaging agents which must be removed by the repair pathway. Many things besides amino acids can bind to RNA; again, this is mostly protein, but can be other things. Amino acids do not spontaneously form large polypeptides. If they did, your cells would not have the nice stockpile of free AAs which they currently possess.
In short, the system is significantly more complex than you think. A college bio course does not get into even 1% of what's going on. My medical molecular genetics course maybe got into another 1% beyond that, and that's being charitable. The mechanisms of cellular replication (which are found even in the simplest bacteria, and to some degree even in viruses) are not a trivial system to build. Please do not offer grand opinions about molecular biology without really understanding how little even the great scientists know about it.
Personally, I find that the complexity of the system leads me to believe in the existence of God. It also leads me to have some ideas about the nature of God, but those are offtopic.
Alik
Um, no, that's when they d-i-e.
Not at all true. Consider your neurons. They do not divide. (There may be some exceptions; the jury is still out.) However, they do not die as long as they continue to receive electrical and chemical stimuli. In fact, most of the cells in your body are in this "senescent" state. An adult has only a few localized populations of continually active dividing cells.
I personally have no idea whether or not this guy is right, but it's a neat theory. As others have pointed out, let him test it; if he's wrong, we can laugh at him, and if he's right we can give him Nobels in Biology and Physics at the same time.
Alik
I am particularly concerned about things like legal notifications being sent to you via e-mail. For very important documents, even postal delivery is not good enough. Some require a return receipt, and some require identity verification (not so much for privacy, but to verify that delivery was made) for delivery. What mechanisms do we have in place, or just have, that can do all this?
Well, for return-receipt, I would suggest something along the lines of the confirmation scheme currently used by many listservs. You sign the initial document via web. They consult your listed contact address with a central key registry and send both a confirmation and some arbitrary bits to that email address. You then sign the arbitrary bits and bounce them back via email. It is now presumable that the order was in fact placed by the person to whom the signing key belongs.
Does this have security flaws? Yes. For example, it remains vulnerable to a man-in-the-middle attack between central key registry and merchant. However, this is a framework created in one minute; a security professional can no doubt generate something a bit more secure.
It can be done using existing protocols and algorithms. I personally would like to see personal keys which are significantly bigger than the 128-bit junk used in the average browser; 2K might hold the line for a few years. (Yes, I know that bigger keys mean more encryption time. How many documents do you actually sign per day? Most people I know don't get past the single-digits.) There are issues in setting up infrastructure, and these must be resolved before you can get me to use such a system, but I think they could even be resolved correctly if people actually bother to think.
Alik
I'm a bit surprised that the big corporations haven't pursued the market for Indian-language computing more aggressively. Many of them have large numbers of bilingual programmers brought over on work visas. Given the sheer size of the potential market in India, I'd expect them to leverage this advantage.
One thing to note (and a possible reason for the lack of translation) is that there is no single "Indian language"; as with Chinese, there are several regional dialects which don't always bear much resemblance to each other. In practice, English ends up getting used a lot of the time as a universal standard. If you walk through Bombay, most of the billboards you see will be in English. The few times I've been there, most of the shopkeepers understood enough English to successfully complete a transaction.
Now, out in the villages, this is not true. On the other hand, I doubt the people in the villages care whether or not they have Internet access just yet. They need sanitation, electricity, and other basic health-improving technology. Once that's in place, then worry about whether or not they can be happy Amazon consumers.
Alik
Crap. Preview box managed to lose my correction to that formula. The correct reaction is
CH2OH + NAD <---> CH2O + NADH
The New Scientist article is baffling because it seems to have things backwards. Our livers (in the unhappy event anyone is foolish enough to consume methanol - not ethanol) in fact convert methanol into formaldehyde and formic acid, not the other way round. It is this formaldehyde and formic acid that account for the toxic effects - notably permanent blindness and destruction of the nervous system and cerebral cortex (not real fun). I know the article waves its hands and says, oh, the process is reversible. Maybe...
Not "maybe". You'll notice that the article also says that they drive the process using NADH. NADH, for those who don't know, is a critical metabolic compound in the pathway from glucose to ATP. (ATP is the primary power source for most cellular reactions that require energy.) The alcohol dehydrogenase reaction, as shown in the article, is:
CH2OH + NAD+ CH2O + NADH
When you have high levels of methanol in your body, the equilibrium is driven to the right. The NADH is promptly removed to the electron-transport system of the mitochondria (or to various synthetic pathways), thus keeping the reaction going in the forward direction. If there's nothing in your system which can remove NADH and you add excess NADH, the equilibrium is driven to the left, transforming formaldehyde to methanol. The other two enzymes in this process are also dehydrogenases, and thus probably also do NAD/NADH conversion, and thus the same rule applies. This is not handwaving, it is basic chemistry.
Now. It is true that there is a question of where the hell one is going to get the necessary NADH. It is also clear that they have no clue yet how this will be done, only some theories. However, the basic chemistry here is perfectly sound. Most metabolic reactions are, in fact, reversible; your body likes to keep its options open.
Alik
For those of us who can't make it, can a transcript of the discussion (with obvious trolls edited out) be posted to Slashdot? Bruce is clearly popular with at least part of the community, so I'd say this is at least as relevant as a regular interview article.
Alik
I think you're quite right about Brooks' current goals and motivations, and I'm equally annoyed by them. I guess it gets more people all fired up about AI/robots, which is nice, but it also tries to tell them that all the problems have simple solutions which don't involve nasty math, which is incorrect.
More fundamentally, a major problem in AI as a field is that people keep trying to make it to human-level AI in one step, rather than clawing their way up the evolutionary ladder with ant-level AI, bee-level AI, lizard-level AI, rodent-level AI, and only then attempting higher mammals.
Why should we do that, though? Logically, all important aspects of intelligent behavior are contained within humanity; therefore, it's easiest to stick to that model. Furthermore, if we work with the lower animals, there's a problem in that they have different sensory modalities, different motion needs, different connection pathways, and so on. (At least, there is if you're doing AI by trying to reverse-engineer the brain, which is my personal favorite method.)
Moreover, what would bee-level or rodent-level or lizard-level mean? I don't actually see a distinction between those levels --- if you have a general solution to the problems of motion planning, target location, and self-preservation, then you can build pretty much any animal you want. (For some animals you'd also need to have manipulation routines in your motion-planning.)
Motion planning et al. aren't really in the critical path for intelligent systems, unless you buy Brooks' thesis that intelligence only exists in embodied form. (I don't. If you made Helen Keller into a quadriplegic, she'd still be intelligent, she'd just go insane.) They are in the critical path for robots, which are generally lumped together with AI because we want AI so we can build bots with it, but the two domains can be separated. Of course, this thread was originally about robots, where such things are very necessary; strangely enough, many robot folks are working on copying the lower animals.
It's certainly not "one step from solution", but at the same time, I think humans are the right level to be working at for the "intelligence" part of things. We have to break the human into subsystems and parcel out the subsystems to different researchers (all of whom will say that their problem would be easy if the other guys would just get theirs solved), but I don't see that we need to solve rats first.
There's hope, though. Game AI developers are struggling to build characters that can survive in tough environments. They're getting better at it. Work in the game AI field is getting to be better than academic AI; the gamers have a real problem and a real market.
I wouldn't go so far as to say that game AI is better. I agree with the previous poster that the game folks haven't exactly made new breakthroughs. They mainly seem to take advantage of the fact that they've got abitrarily perfect sensing, motion, aiming, communication, coordination, and so on. Beyond that, game AI still seems to mostly be preprogrammed formations and decision trees. (On the other hand, many real soldiers work that way.) Even Deep Blue wasn't a particular breakthrough in technique --- they took the standard search tree chess algorithm, tuned it for Kasparov's playing style, and threw Moore's Law at the problem. Blue Gene will be something similar for the protein-folding problem. (Not that I care --- protein-folding has to be figured out somehow, and whatever works benefits everyone.)
I think a good goal for game AI is just-too-hard AI: a system which is always a little bit better than you, and thus keeps forcing you to improve on your own. Especially for things like chess, a system like that (with skill implemented as a limit on search tree depth and possibly multiple search algorithms) would make an excellent trainer. Sadly, I doubt it could teach me to play good Quake, because as far as I can see human Quake players don't follow patterns.
Alik
The email in question said something to the effect of "The AC's karma is currently 1972. Way not negative." You are welcome to doubt and ask Rob to rerun the calculation; perhaps he accidentally inverted a sign somewhere. I choose to believe he did it right.
Alik
can you reproduce a human baby exactly? would you want to? maybe if people just valued individuality and life this whole 'reproducibility' thing would not matter
That's one thing when speaking of a human life (and there are still cases when I would want to duplicate a baby), quite another when speaking of a scientific project. The Cog project is a scientific research endeavour. One of the criteria for good science is reproducibility of results. If you can't do it again, it's not a particularly big breakthrough.
Alik
Yes, humans might be that advanced as to come up with new ideas, by contemplating a problem long enough, but even for us (limited) random behaviour can bring unexpected rewards. I don't think the first people to discover burned (i.e. roasted) meat tastes pleasant or the primate noticing that using rocks as hammers opens things, did so by reasoning.
I agree, but I wouldn't define such things as random behavior. I would define them more as accidental discoveries. For example, cooking probably didn't come about because a human randomly decided to throw meat into a fire; it more likely occured by accidentally dropping meat into the fire, or by finding an animal which had somehow been caught in a fire. The randomness is on the part of the world, not of the agent.
Intentional random behavior in the hopes of discovering something new and interesting seems to be to be the basic principle behind genetic algorithms. GAs are neat, but personally I find the stuff they produce to be pretty ugly and non-robust. (Of course, so were some early robot programs.) GAs are another technology that I think gets a bit too much hype; I can see a bitstring evolving into "Hello, world!", but I can't see one evolving into a stable Mozilla.
Alik
Just because you don't know the emotion algorithm in humans doesn't mean it doesn't exist.
That's fair enough. However, I would say that if there is an algorithm, it is certainly not an obvious one.
Let us not confuse lack of comprehension with self-involved flattery. We are not that complex of beings.
Here I start to disagree. I believe that we are fairly complex beings, at least for any human definition of complexity. "Complex", in common usage, translates to "Damn hard to do and/or understand." I think it's perfectly reasonable to state that the human mind is damn hard to build and/or understand. What systems do you see that are so much more complex than the body? The only ones I can think of are those in which humans are components, such as social systems.
Everything we are is matter, and therefore we have a protocol in order to be a living being.
I disagree with this statement, or at least with what I believe you mean by this statement. If you are saying that mind can be solely reduced to the motion of particles within the space defined by your brain, I disagree. I believe that there is such a thing as a soul. I also believe that the soul can be detected and measured, but that we do not yet know how to do so. This means that the soul is, in some sense, matter (or energy, which is the same thing), and therefore we are in fact purely material. However, in general, when someone says "All we are is matter," they mean it as a denial of the soul.
You assume our emotions aren't predictable simply becuase we are not smart enough to be able to, yet. Our algorithm is obviously far more advanced then:
Actually, it's a bit more than that. Let's assume for a second that you're right and that our minds are nothing more than the known matter within our brains and bodies. You cannot accurately predict what the behavior of that matter will be. The actions of the brain depend on the motions of single ions, which in turn depend on the motions of charged particles, which are fundamentally unpredictable by the Heisenberg principle. This is a good thing --- it provides an empirical proof of something approximating free will.
But, you can keep believing what you want because we won't figure it out in your lifetime.
That's a pretty bold assertion with no backup, and I suspect it of being said just to incite and inflame. How can you predict what science will do? How can you predict how long I might live? Personally, I intend to see the year 2200, and hopefully many years beyond that.
Alik
As much as it should be adressed (to make robotics are "serious" since ;-> IMHO it is part of the question, as "intelligent behaviour" has to contain some sort of randomness and unpredictability.
I'm not quite able to parse the first part of your sentence (especially because of an unclosed paren), but I can handle the second part, and I'm not sure I agree. I don't think intelligent behavior has to be random or unpredictable. Humans are not random. Humans are, to some degree, unpredictable. However, much of that comes down to how we're designed; another large factor is emotion.
Our first attempts at AI and semi-intelligent robots will probably not have emotion, just because we have even less clue about how emotion works than we do vision. They may make expressions (this is something the Kismet project at MIT is apparently doing well) for cuteness factor, but I doubt they'll have anything we call "real" emotion. I believe that their intelligence will be largely algorithmic in nature, especially the motion-planning parts of it. Therefore, their movements will be totally or almost-totally predictable, as will their responses to stimuli.
I can see an argument that some randomness will be built in because humans would be uncomfortable with something that always moved in the exact same way, but that does not imply that intelligence requires randomness.
Alik
BTW, for anyone interested there is the COG project at MIT, doing research into a humanoid robot, it is VERY impressive.
"Any sufficiently advanced technology is indistinguishable from a rigged demo." --Anon of Ibid (or somebody)
Seriously, though, while I'm not accusing Rodney Brooks & Co. of being frauds, I do think they're getting credit for a lot more than they've really done. Reactive control is wonderful for simple systems. In the MIT robot competitions, the second-place robot is generally some simple reactive design. (At Dartmouth, the first-place robot was, but the last Dartmouth robot competition had a bot construction/program time of about six weeks.)
However, the robot that wins out is generally not using a Cog-like architecture; it's usually one of the designs that uses a more artificial method of intelligence and navigation. The problem with reactive control is that it's easy to build and debug for simple systems, but as you add more functions and layers the connections and interactions become too much for even an MIT mind to handle. Brooks even says as much in one of his early papers on subsumptionism.
There's also Brooks' admission in his own FAQ that even the simple ant robots they've built are utterly unreproducible. IMHO, it's not that great if you can build something once but can't tell other people how to reproduce it.
This isn't meant to be a whole bunch of doomsaying about reactive design --- I think it's appropriate in some circumstances, and I've used it myself when I needed a simple barely-effective hack. It's just that sometimes it feels like the Cog Shop and their academic relations would like people to believe that they're the ultimate solution to all of robotics, and I don't think that's going to happen. It's too hard to use their methods in a general and structured manner to build systems that have anything resembling guaranteed behavior.
Alik