For a computer, playing chess is just crunching numbers. Thanks to Moore's law, the top chess players will soon consistently fall to the machines.
However this is not the case with the game of go. First the search space is much bigger. This should hold off the machines for a while. But most importantly, we haven't found a good evaluation function. This function is what the game engine uses to choose the "best" move among the ones it evaluates. So without a good evaluation function, no matter how many billion moves the machine can evaluate per second: it will have no idea which one to choose.
And it would be a catastrophe waiting to happen, when (not if) it snaps and rains megatons of carbon cable down upon the ground below.
I'll bite.
The ribbon would weight 7.5 kg/km (yes, that's per kilometer). So the whole ribbon with its 100,000 km would weight 750,000 kg. Of course only the portion below the severed point would fall back. If that happens at an altitude of 1,000 km (region of space debris), only a mere 7,500 kg would fall back to Earth.
7,500 kg...
1 megaton = 1,000,000 t = 1,000,000,000 kg
Your apocaliptic image of "rains megatons of carbon cable" is off by 6 orders of magnitude. Not to mention that with its low density, the ribbon would not plummet down, but would probably be picked up by winds and fall quietly, much like a piece of cloth.
Just watch out for the climber. Oh here it comes! "INCOMING!"
"Technically it's feasible," said Robert Cassanova, director of the NASA Institute for Advanced Concepts.
[...]
The key to the concept's feasibility lies in the material that will be used to construct the ribbon between the Earth and outer space. Nanotubes are essentially sheets of graphite -- a lattice of carbon -- seamlessly rolled into long tubes that are mere nanometers in diameter. These are 100 times as strong as steel, but much lighter.
"Carbon nanotubes are rapidly developing," Cassanova said. "They are not long enough to stretch from Earth's surface to 62,000 miles, but there are a number of organizations working on that now."
If you've ever installed a Linux distribution, you will immediately note the number of third-party libraries and applications installed on a 'base' system. This is frustrating for me, because for the most part I may not want all those extra applications installed, because it clutters up the system, and there may be various vulnerabilities present that I'll be open to.
You should give Linux From Scratch a try. You will build your own Linux system, installing each component manually. No clutter, just what you need, and compiled the way you want it. It is a very good learning experience too.
Ah, we have the opportunity to restart an old flame war!;^)
<flame> The real hackers were not working on Amigas but on Ataris: you had no blitter chip and such on the Atari ST, you had to do everything "by hand". so making demos on an Atari, equivalent in quality to the Amiga ones (and they were), was much more of a challenge! </flame>
LOL. Sorry the toaster-microwave slot is already taken;-) Actually the main problem is that you usually get a single IP. Sometimes it's not even static (depends on which ISP you use).
So while these FTTH connections are great for personal or small offices use using NAT, you wouldn't be able to start you personal datacenter on them.
Note: some ISP do provide mutliple IPs on FTTH connections. But usually, their backbone link sucks. I've seen one offering 8 static IPs for the same price I'm paying now, but it has a single 100 Mbps link to an Internet exchange point. 100 Mbps? Then how I am supposed to get 100 Mbps myself if they share that among their customers? What if that link goes down? Thank you but I'd rather have my single static IP, with redundant backbone connections and a total close to 5 Gbps of bandwidth if I remember well.:-)
I was a bit worried about that when I applied and was wondering what bandwidth I would actually get. The nice thing is: it does work as advertised! See my other post here.
Note: the actual bandwidth you get depends on which ISP you choose (the 100Mbps connection is to your ISP, not to the Internet directly). I chose mine carefully:)
You are describing Hong Kong. Most houses in Tokyo are actually 2 or 3 stories. Yes, they are all next to each other, which results in a very high population density. But fiber still has to be laid to every house. And it is being done so.
Sorry I wasn't clear enough there: 5 years ago, ISDN was just starting to spread IN JAPAN. So thank you, you help me make my point: Japan went from being very late, regarding end users Internet connectivity, to now being at the top. And that, in a 4 years span (100 Mbps fiber has been running for a year now. See my other post here).
Now regarding my question (yep, I'm serious, that's not a troll): population density is indeed a strong factor against DSL: you have to be within a few km of the telco switch. And indeed everybody seems to be answering that. But that's not a problem with fiber. And as the parent article says, there IS plenty of dark fiber already laid in the US, just waiting to be used. So the question remains...
These FTTH services started about a year ago, with coverage only in central Tokyo and Yokohama. A few companies are competing on that market, mainly NTT (used to be the Japanese telephone monopoly) and Usen (Cable TV provider). This competition probably explains the rapid expansion of FTTH: in less than a year, most of Japan is covered! (Japan population, not land. But still, I'm impressed!).
For some more info, here is a link to NTT's english description of their service. You have to add ISP fees to the numbers on this page (about 4000 yens/month, depending on the ISP). I did the math for you:
For a family type connection (you share the bandwidth with a few neighbours), that'd cost you about $65/month.
And for a basic type connection (100 Mbps just for you! Yes, in your kitchen if you insist;-) ), that's about $100/month.
Note that the bandwidth isn't garanteed (Does any provider garantee bandwith anyway?), but, at least in our case, it works as advertised: we achieve top download speed of over 90 Mbps, and I routinely download the 2 or 3 Gb of some Linux distribution in 20 or 30 minutes (that is, when the mirror can handle it. Ah, feels great to know that your local connection isn't the bottle neck!).
Now enough with the free advertising, on to the thinking.
You are right about phone calls: they are still quite expensive. Even local calls are charged by the minute. Why? Because NTT had their monopoly for years and could set the prices very high. Now that the monopoly is broken, the competitors need to set their price just a little below NTT's to attract people. Why would they go below that and make less money? Noone has a real incentive to really make phone calls cheap: everyone keeps the prices high, everyone makes a lot of money, everyone is happy! (Yeah, "customers" ain't included in that "everyone"...)
Now what about this affordable 100Mbps Internet access? Did the greedy corporations suddently decide to become humanists and work for the good of mankind, providing us with affordable broadband? Of course not! As for any corporation, there is only one goal: make money. And they are doing good: if I was still on a 56 kbps modem, I wouldn't be shedding 100 bucks a month to them!$20 maybe, and I'd be bitching that their service is crap and overpriced...
But now I do shed 100 bucks a month! They get 100 bucks, they are happy. I can download Debian in 30 minutes, I'm happy. Everyone's happy!
Note that this time, the customer IS included in "everyone". The main difference with the previous scenario is: the first scenario has a monopoly, while the second has competition.
Hmmm... Didn't plan to finish this posting like that, but that's a good conclusion: Competition GOOD! Monopolies BAD!
Japan has been laying dark fiber for over 10 years now. Many were laughing at them during that time, but today, we have a 100 Mbps fiber internet connection coming right into our kitchen! For something around 100$ a month. Ok, not super cheap yet, but affordable, specially if you share it.
Five years ago, the top for end users was still 56 kbps modems, that was just the begining of ISDN. Pretty impressive evolution.
Now question: if dark fiber is there, why is it that you still can't get decent DSL internet connection in the US? What's hindering the development of broadband there?
When an hour of downtime would cost you millions of dollars, no question about it: you get a mainframe.
For the ones who don't read the article, a quick excerpt so you know what kind of availability we are talking about:
"[...] today's [mainframe] systems [are] so reliable that it is extremely rare to hear of any hardware related system outage. There is such an extremely high level of redundancy and error checking in these systems that there are very few scenarios, short of a Vogon Constructor fleet flying through your datacenter, which can cause a system outage. Each CPU die contains two complete execution pipelines that execute each instruction simultaneously. If the results of the two pipelines are not identical, the CPU state is regressed, and the instruction retried. If the retry again fails, the original CPU state is saved, and a spare CPU is activated and loaded with the saved state data. This CPU now resumes the work that was being performed by the failed chip. Memory chips, memory busses, I/O channels, power supplies, etc. all are either redundant in design, or have corresponding spares which can be can be put into use dynamically. Some of these failures may cause some marginal loss in performance, but they will not cause the failure of any unit of work in the system."
I don't know about you guys, but I usually don't look at the remote control when zapping between channels, changing the volume. Expect for the most exotic functions, I never need to look down and find the key. I just "touch type" on the remote control.
With this new device, that wouldn't be possible. Ok, you can have several remote controls integrated into one. But we already have that with normal remote controls...
So, really, apart from the geek fractor, this thing just sounds like a pain to use. Imagine a keyboard without keys! Oh, but wait...
Actually, the exponentially larger sets involved are not the main issue. Sure, it makes finding a good move a longer task, but remember Moore's law: computer power grows exponentially too. So, if that was the only problem, we could guaranty that computers would eventually become go experts.
However, the main problem is deeper. It is at the root of the combinatorial game theory that sits behind Fritz' victories: the evaluation function. As mentioned in the article you link to:
The problems with computer go start with the size of the tree [...] The real problems begin, though, with the evaluation function. Computers have great difficulty deciding which is the better of two go positions, and cannot even reliably say who is the winner when the game is over.
Computers are just number crunching machines. If you can't feed them with numbers, them will do you no good. So that's what this evaluation function does: it takes a given board situation, and evaluates its value to the player. These values are computed for all the leaves of the explored moves tree, and are used to prune this tree down. Finally, you are left with the best next move. Therefore, 'best' depends not just on the size of the tree, but also on the quality of the evaluation function.
The bottom line is:
A sub-quantum galactic-class megascale supercomputer that could compute '42' before it is destroyed to make way for a bypass, without a good go board evaluation function, would still suck at go.
And that is exactly what the problem is. While it is pretty easy to say (compute!) who has an edge on a chess board, we still haven't found a good computational way to do judge the value of a go board situation. Of course experts go players DO have a feeling of a better situation, but that's exactly it: a matter of feelings, not of equations. Until we have put these go feelings into equations, whatever the processing power, computers will just suck at go.
A device that lets you exchange business cards through a simple handshake...
invented in Japan, where indeed business cards exchange is a national sport...
but where direct physical contact is always avoided (Japanese people bow at each other, remember?).
The article basically asks "Should we work on interop for CMS systems?" (and answers yes). But since 100% interop is way out of reach (everyone who developed with different CMS systems will agree), and the only way is really to tackle the problem one baby step at a time, there is a missing question:
What baby steps should the OSCOM Interop project work on first?
The first steps were already taken with WebDAV and XML-RPC. What should be next?
Here is a story I heard a while ago that fits well here. I haven't been able to verify it yet, so any additional info would be appreciated. Here is how it goes:
Back in the days when Sony was a no-name outside Japan, two Japanese guys were sent to Europe to prepare the market for the entry of their company's products. Not really knowing how to do, and maybe because they liked the fine german brews, they came up with the following plan: for 6 months, all they did was tour the bars of Germany asking for a "Sony please?".
The story doesn't say if they were ever served a sony cocktail (a sunny maybe?), but their campaign worked beyond their expectations, with word-of-mouth spreading questions about these "Japanese dudes asking for a sonee... --A what? --A sonee... --Hmmm, a... sonee... Never heard of that one. I'll ask around."
When Sony started selling their products their, everybody had heard the name, and noone had a clue about what it meant! I know many here will dislike this marketing technique, but the thing is: it works. And in marketing, it's all that counts.
From the proceedings of the Royal Society in which the original paper was published (emphasis is mine):
Counterfactual computations (p. 1175)
By G Mitchison & R Jozsa
One of the more remarkable predictions of quantum mechanics is that it should be possible to detect the presence of an object, using a photon as a probe, without any interaction occurring between the photon and the object. Richard Jozsa recently showed that one could extend this idea by replacing the object with a computer. One can then discover what the outcome of the computation would be without running the computer! We discuss the meaning of such ecounterfactual computationsf, and show that there are some limits on the information that can be obtained from them.
I guess everything is in these "limits"... Has anyone had access to the full paper and could tell us more about them?
Nope, the original is right: B4I4QRU/16. Remember how you read a fraction "a/b"... IANANES (I am not a native english speaker), but I think it is something like "a over b", right?
Slashdot Mirror
in the game of go.
For a computer, playing chess is just crunching numbers. Thanks to Moore's law, the top chess players will soon consistently fall to the machines.
However this is not the case with the game of go. First the search space is much bigger. This should hold off the machines for a while. But most importantly, we haven't found a good evaluation function. This function is what the game engine uses to choose the "best" move among the ones it evaluates. So without a good evaluation function, no matter how many billion moves the machine can evaluate per second: it will have no idea which one to choose.
I'll bite.
The ribbon would weight 7.5 kg/km (yes, that's per kilometer). So the whole ribbon with its 100,000 km would weight 750,000 kg. Of course only the portion below the severed point would fall back. If that happens at an altitude of 1,000 km (region of space debris), only a mere 7,500 kg would fall back to Earth.
7,500 kg...
1 megaton = 1,000,000 t = 1,000,000,000 kg
Your apocaliptic image of "rains megatons of carbon cable" is off by 6 orders of magnitude. Not to mention that with its low density, the ribbon would not plummet down, but would probably be picked up by winds and fall quietly, much like a piece of cloth.
Just watch out for the climber. Oh here it comes! "INCOMING!"
According to this article mentioned earlier on Slashdot:
You should give Linux From Scratch a try. You will build your own Linux system, installing each component manually. No clutter, just what you need, and compiled the way you want it. It is a very good learning experience too.
Ah, we have the opportunity to restart an old flame war!
<flame>
The real hackers were not working on Amigas but on Ataris: you had no blitter chip and such on the Atari ST, you had to do everything "by hand". so making demos on an Atari, equivalent in quality to the Amiga ones (and they were), was much more of a challenge!
</flame>
There are over 24,361,450 businesses in the US
Ah yes, I forgot:
"According to the MPAA, there are over 65,744,682 businesses in the US. They actually found 24,361,450 but some of them were big corporations."
There are over 24,361,450 businesses in the US (Source here)
...
...
:-) Bottom line is: don't trust numbers, you can make them say whatever you want!
So let's see. If:
o every US business was doing 1 mass mailing a year;
o that day would be picked at random;
o your email ended on 2% of these mass mailing lists;
The you would be getting:
24,361,450 mailings / 365 days * 2%
(launching calc...)
= 1335 spam per day
Hey that was close enough!
PS: yes, I made up the assumptions to match the result
LOL. Sorry the toaster-microwave slot is already taken ;-) Actually the main problem is that you usually get a single IP. Sometimes it's not even static (depends on which ISP you use).
:-)
So while these FTTH connections are great for personal or small offices use using NAT, you wouldn't be able to start you personal datacenter on them.
Note: some ISP do provide mutliple IPs on FTTH connections. But usually, their backbone link sucks. I've seen one offering 8 static IPs for the same price I'm paying now, but it has a single 100 Mbps link to an Internet exchange point. 100 Mbps? Then how I am supposed to get 100 Mbps myself if they share that among their customers? What if that link goes down? Thank you but I'd rather have my single static IP, with redundant backbone connections and a total close to 5 Gbps of bandwidth if I remember well.
I was a bit worried about that when I applied and was wondering what bandwidth I would actually get. The nice thing is: it does work as advertised! See my other post
:)
here.
Note: the actual bandwidth you get depends on which ISP you choose (the 100Mbps connection is to your ISP, not to the Internet directly). I chose mine carefully
You are describing Hong Kong. Most houses in Tokyo are actually 2 or 3 stories. Yes, they are all next to each other, which results in a very high population density. But fiber still has to be laid to every house. And it is being done so.
Sorry I wasn't clear enough there: 5 years ago, ISDN was just starting to spread IN JAPAN. So thank you, you help me make my point: Japan went from being very late, regarding end users Internet connectivity, to now being at the top. And that, in a 4 years span (100 Mbps fiber has been running for a year now. See my other post
here).
Now regarding my question (yep, I'm serious, that's not a troll): population density is indeed a strong factor against DSL: you have to be within a few km of the telco switch. And indeed everybody seems to be answering that. But that's not a problem with fiber. And as the parent article says, there IS plenty of dark fiber already laid in the US, just waiting to be used. So the question remains...
These FTTH services started about a year ago, with coverage only in central Tokyo and Yokohama. A few companies are competing on that market, mainly NTT (used to be the Japanese telephone monopoly) and Usen (Cable TV provider). This competition probably explains the rapid expansion of FTTH: in less than a year, most of Japan is covered! (Japan population, not land. But still, I'm impressed!).
;-) ), that's about $100/month.
For some more info, here is a link to
NTT's english description of their service.
You have to add ISP fees to the numbers on this page (about 4000 yens/month, depending on the ISP). I did the math for you:
For a family type connection (you share the bandwidth with a few neighbours), that'd cost you about $65/month.
And for a basic type connection (100 Mbps just for you! Yes, in your kitchen if you insist
Note that the bandwidth isn't garanteed (Does any provider garantee bandwith anyway?), but, at least in our case, it works as advertised: we achieve top download speed of over 90 Mbps, and I routinely download the 2 or 3 Gb of some Linux distribution in 20 or 30 minutes (that is, when the mirror can handle it. Ah, feels great to know that your local connection isn't the bottle neck!).
Now enough with the free advertising, on to the thinking.
You are right about phone calls: they are still quite expensive. Even local calls are charged by the minute. Why? Because NTT had their monopoly for years and could set the prices very high. Now that the monopoly is broken, the competitors need to set their price just a little below NTT's to attract people. Why would they go below that and make less money? Noone has a real incentive to really make phone calls cheap: everyone keeps the prices high, everyone makes a lot of money, everyone is happy! (Yeah, "customers" ain't included in that "everyone"...)
Now what about this affordable 100Mbps Internet access? Did the greedy corporations suddently decide to become humanists and work for the good of mankind, providing us with affordable broadband? Of course not! As for any corporation, there is only one goal: make money. And they are doing good: if I was still on a 56 kbps modem, I wouldn't be shedding 100 bucks a month to them!$20 maybe, and I'd be bitching that their service is crap and overpriced...
But now I do shed 100 bucks a month! They get 100 bucks, they are happy. I can download Debian in 30 minutes, I'm happy. Everyone's happy!
Note that this time, the customer IS included in "everyone". The main difference with the previous scenario is: the first scenario has a monopoly, while the second has competition.
Hmmm... Didn't plan to finish this posting like that, but that's a good conclusion: Competition GOOD! Monopolies BAD!
May should have phrased that more clearly, something like:
"If the technology is available and sometimes already deployed, what's hindering the development of broadband internet connections in the US?"
Japan has been laying dark fiber for over 10 years now. Many were laughing at them during that time, but today, we have a 100 Mbps fiber internet connection coming right into our kitchen! For something around 100$ a month. Ok, not super cheap yet, but affordable, specially if you share it.
Five years ago, the top for end users was still 56 kbps modems, that was just the begining of ISDN. Pretty impressive evolution.
Now question: if dark fiber is there, why is it that you still can't get decent DSL internet connection in the US? What's hindering the development of broadband there?
Can you say "banks"?
When an hour of downtime would cost you millions of dollars, no question about it: you get a mainframe.
For the ones who don't read the article, a quick excerpt so you know what kind of availability we are talking about:
"[...] today's [mainframe] systems [are] so reliable that it is extremely rare to hear of any hardware related system outage. There is such an extremely high level of redundancy and error checking in these systems that there are very few scenarios, short of a Vogon Constructor fleet flying through your datacenter, which can cause a system outage. Each CPU die contains two complete execution pipelines that execute each instruction simultaneously. If the results of the two pipelines are not identical, the CPU state is regressed, and the instruction retried. If the retry again fails, the original CPU state is saved, and a spare CPU is activated and loaded with the saved state data. This CPU now resumes the work that was being performed by the failed chip. Memory chips, memory busses, I/O channels, power supplies, etc. all are either redundant in design, or have corresponding spares which can be can be put into use dynamically. Some of these failures may cause some marginal loss in performance, but they will not cause the failure of any unit of work in the system."
I don't know about you guys, but I usually don't look at the remote control when zapping between channels, changing the volume. Expect for the most exotic functions, I never need to look down and find the key. I just "touch type" on the remote control.
With this new device, that wouldn't be possible. Ok, you can have several remote controls integrated into one. But we already have that with normal remote controls...
So, really, apart from the geek fractor, this thing just sounds like a pain to use. Imagine a keyboard without keys! Oh, but wait...
You mean like Popular Power tried (and failed)to do? Check their old site to see what they used to propose.
Looks like selling CPU cycles is not a lucrative business...
Actually, the exponentially larger sets involved are not the main issue. Sure, it makes finding a good move a longer task, but remember Moore's law: computer power grows exponentially too. So, if that was the only problem, we could guaranty that computers would eventually become go experts.
However, the main problem is deeper. It is at the root of the combinatorial game theory that sits behind Fritz' victories: the evaluation function. As mentioned in the article you link to:
Computers are just number crunching machines. If you can't feed them with numbers, them will do you no good. So that's what this evaluation function does: it takes a given board situation, and evaluates its value to the player. These values are computed for all the leaves of the explored moves tree, and are used to prune this tree down. Finally, you are left with the best next move. Therefore, 'best' depends not just on the size of the tree, but also on the quality of the evaluation function.
The bottom line is:
And that is exactly what the problem is. While it is pretty easy to say (compute!) who has an edge on a chess board, we still haven't found a good computational way to do judge the value of a go board situation. Of course experts go players DO have a feeling of a better situation, but that's exactly it: a matter of feelings, not of equations. Until we have put these go feelings into equations, whatever the processing power, computers will just suck at go.
All you need is... Progress Quest
Sorry, but I'm ROTFLASTC...
A device that lets you exchange business cards through a simple handshake...
invented in Japan, where indeed business cards exchange is a national sport...
but where direct physical contact is always avoided (Japanese people bow at each other, remember?).
Life is wonderful.
The article basically asks "Should we work on interop for CMS systems?" (and answers yes). But since 100% interop is way out of reach (everyone who developed with different CMS systems will agree), and the only way is really to tackle the problem one baby step at a time, there is a missing question:
What baby steps should the OSCOM Interop project work on first?
The first steps were already taken with WebDAV and XML-RPC. What should be next?
A couple of years ago, a meteorite hit a house in Japan (article in Japanese). Went straight through the roof.
Now ok, the rock that hit that girl's foot is smaller, but it's still hard to believe it just bounced off her shoe...
Here is a story I heard a while ago that fits well here. I haven't been able to verify it yet, so any additional info would be appreciated. Here is how it goes:
Back in the days when Sony was a no-name outside Japan, two Japanese guys were sent to Europe to prepare the market for the entry of their company's products. Not really knowing how to do, and maybe because they liked the fine german brews, they came up with the following plan: for 6 months, all they did was tour the bars of Germany asking for a "Sony please?".
The story doesn't say if they were ever served a sony cocktail (a sunny maybe?), but their campaign worked beyond their expectations, with word-of-mouth spreading questions about these "Japanese dudes asking for a sonee... --A what? --A sonee... --Hmmm, a... sonee... Never heard of that one. I'll ask around."
When Sony started selling their products their, everybody had heard the name, and noone had a clue about what it meant! I know many here will dislike this marketing technique, but the thing is: it works. And in marketing, it's all that counts.
"Are you greater than 16" ?!
Nope, the original is right: B4I4QRU/16. Remember how you read a fraction "a/b"... IANANES (I am not a native english speaker), but I think it is something like "a over b", right?
--LHOOQ (tip: that one is in French...)