Here's why "10x as accurate as human" is meaningless: Statistical filters are trained by human input. If the human input is only 99,84% accurate, then you cannot trust the filter to do any better.
That goes with a caveat: If the human classification mistakes are random, then it is possible for the filter to do better. But if the mistakes are systematic, e.g. if you trust all messages containing the text "slashdot.org" to be ham, then that dire mistake in the input will carry over to the output. Garbage in, garbage out.
You owe a debt of gratitude to the people who throw rocks through your windows. If it weren't for them, you wouldn't bother to board over your windows, and then you'd REALLY be screwed when someone breaks your windows to pipe poison gas through.
What will they do after they've finished with Y and Z, they'll have no more letters of the alphabet then. It'll be a disaster
Then Norway will rise to world domination with the AE Window System.
The first thing that will then happen is that Slashcode will be eradicated; because that piece of Perl junk won't let me use a proper Æ in comments.
The Danish name for a fruit machine translates into a one-armed thief. Now show me a slot machine that doesn't steal the gamblers money, that would be news.
So... in this macho atmosphere, reeking of testosterone... the punishment for not being that the bug meet is that... YOU DON'T HAVE TO FIX YOUR BUGS UNTIL THE NEXT MAJOR RELEASE?????????
It seems the MS developers consider it a punishment to ship a product with a bug in the parts they are responsible for, and you consider it a reward. What does that say about the MS developers, and what does that say about you?
This patent seems valid - it is certainly nontrivial . I hope Microsoft will have spectacular succes in defending it, and that they will charge anyone wanting to use it a prohibitively huge ransom. Oh, and Microsoft, if you you'd just sue anyone trying to market competing DRM technology for infringement, please?
This is patents doing what they do best: Preventing technology from being used!
/A
Re:Uhmm, sorry! Lot's of prior art here ;-)
on
MS Palladium Patent
·
· Score: 1
"The computerized method of claim 1, wherein protecting the rights-managed data comprises: refusing to load the untrusted program into memory."
manyoso: Hmmm. Seems to me that this 'art' has been around since the beginning of Unix. Hell, Microsoft has been providing a form of this 'art' with NT and 2000 for quite sometime. It's called permissions!
The difference being that you will not have root rights on your own hardware.
But HTML doesn't need to be "heavy"! You can make skinny pages, you can make heavy pages, but no one forces you to make heavy pages if you don't want to!
Right; but HTML is heavy in another sense: the code size and processing power required to parse and render HTML is huge. Huge, that is, for the low-power cpu in a mobile device.
To use HTML, a mobile device would have to limit itself to some arbitrary subset, which is bad because every device manufacturer would choose a different subset. To standardize on an XHTML subset actually sounds like a good idea.
There are web browsers that run on Commodore 64 - all the way up from TCP/IP stack to HTML to formatting... That's a thin client for you!
I'm sure those web browsers are great as long as you stick to the particular HTML subset that they can handle well. And I'm sure they suck when you don't.
Visual Basic is a very poor language... VB's string concatenation is O(n^2)!
Ouch! They must have worked very hard at Redmond to make it *that* bad, given that the obvious string concat implementation is O(n). That would make naive concatentation of N strings K characters each O(K N^3).
But that is a language implementation problem. Not a language problem per se. Nothing's to stop the KBasic team from doing better. They could do a lot better, actually: with a little thought and a little exponential overallocation in the string implementation, naive concatenation of N K-character strings would be only O(K N lg(K N)).
Hmmm, and where can I get this wonderful managed security? Why look, Bruce himself sells it! What a happy surprise...
Of course. He
Analyzed the problem and realized that managed security was the way to go.
Created a product that does just this.
Wrote an article based on his analysis.
Ok, so he could have chosen to do step 2 and 3 in reverse order. (Actually, maybe he did and we just didn't notice?) But that doesn't invalidate
the analysis. Nor the product.
If you're lucky to have a holiday that lasts several years, that is.
Hit the throttle, accelerate at 10 m/s2 the first day, decelerate at the same rate the second day, and you're there, and you even had gravity all the way. Why would you want to spend years on Mars?:-)
Sure, with today's staged rockets and slingshot technology the voyage alone would take years. Which is exactly why I wouldn't go today.
What do you mean by "more efficient" ways of space travel? Let's investigate your ideas one by one :
Whoops. My title "Build the space infrastructure first" was slightly off-key. It should have said "Invent the space infrastructure first". It wasn't that I have a lot of brilliant ideas as to how space travel should really be done, and every single one of the specific ideas that I hintet at may well be useless. But if these ideas don't pan out then we need to get some other great idea, because today we really don't have a clue on how to do space travel in a reasonably efficient manner.
The fundamental problem of how to get materials, mainly fuel, free of earth's gravitational pull, will not be solved by going to Pluto. Such a project would just use the same old big multi-stage rocket, which is proven technology. It is simpler to buy a huge fuel tank and fill it up than it is to develop some future technology that we haven't even begun to imagine yet. But we need the latter if it's ever going to be more than the odd one-off giga-dollar expidition.
Repair Facilities : In space, you don't "repair" something : you replace them. It's WAAY cheaper to be redundant in components than to build elaborate facilities to repair things..
Quite right. But again, it doesn't scale. So we'll just have to figure out how to build simple, generic repair facilities instead of elaborate, bulky, a-tool-for-every-job facilities.
A better challenge than going to Pluto would be this: Build a self-sustaining space station. In orbit, on the Moon, on Mars, I don't care. Any which way it is a formidable challenge, but in return we would learn a lot, not just about space, but also about ourselves, ecosystems, life.
Rather than spending the big $$$ on going into space now, they should be spent on finding more efficient means of space travel.
They should be researching propulsion systems, building waystations in orbit, building lunar refueling and repair facilities, gas^H^H^Hwater stations for fusion reactors etc.
In 100 years, when we have a reasonable handle on these things, we can begin sending probes into deep space, taking holiday trips to Mars and much much more, and we will learn infinitely more than we would learn from a Pluto mission today. To attempt these things before we have developed decent space travel logistics, is just a waste of resources.
OK, so you and I will not live to see the the results of such a long term plan. But then space exploration is inherently long term.
But for people who don't work with computers all day long, saying something like "Enter a check," "Open a new document," or "Send an e-mail" will do just fine.
"Enter a check", huh?
The chess 'bot would wake up and tell you
Sorry Dave, I can't let you do that. Chess rules specifically forbid entering a check.
Seriously, ambiguity is a big problem with natural-language interfaces. How is the computer to know if you wanted to do something involving your checking account or if you want to add an integrity constraint to a computer program you are writing? Or if you are playing a chess game and trying to make an illegal move...
Natural language comprehension is an AI-complete problem. That doesn't mean useful approximations can't be done. But they only work if you keep commands simple and adhere to a computer-friendly style of expression. With requests of any complexety the risk of a misunderstanding is too great to trust interface to do anything that can't easily be undone. In this sense it is similar to DWIM interfaces (follow the link for a good anecdote).
No doubt natural language interfaces will find its niches, and someday many people may even be using natural language interfaces exclusively. But when these people need to do something more complex or risky, they will need to turn to a hacker who masters some other arcane but concise command language.
Disclosure causes security holes to get taken advantage of.
Full disclosure is suboptimal because people have better things to do with their time than upgrade and patch software. No disclosure is suboptimal because there is no pressure on software vendors to fix holes.
Full disclosure works well with stable software. Eventually bugs get fixed and the continuous public review will have created a secure product which can be used for years and years.
It is with rapidly changing software there is a problem. And in these days where "internet time" is a valid excuse for anything, rapidly changing software is what we have lots of. For this, we need a good idea for how to strike a balance between the two extremes, full disclosure and full secrecy.
One idea is to have a "waiting period". When someones discovers a security problem they inform the vendor but wait some about of time before informing the public, say 1 month. That way not only will a fix be ready place when the problem is publicized, but frequent upgraders may already have the patched software running. The software vendor, knowing that the problem will be publicized at the end of the waiting period, still have an incentive to get it fixed.
Of course this doesn't help the masses with years-old software. Someone please get an even better idea!
Slashdot Mirror
Here's why "10x as accurate as human" is meaningless: Statistical filters are trained by human input. If the human input is only 99,84% accurate, then you cannot trust the filter to do any better.
That goes with a caveat: If the human classification mistakes are random, then it is possible for the filter to do better. But if the mistakes are systematic, e.g. if you trust all messages containing the text "slashdot.org" to be ham, then that dire mistake in the input will carry over to the output. Garbage in, garbage out.
/A
You owe a debt of gratitude to the people who throw rocks through your windows. If it weren't for them, you wouldn't bother to board over your windows, and then you'd REALLY be screwed when someone breaks your windows to pipe poison gas through.
Poor parents => worse nutrition => shorter children
+
Poor parents => worse education => poor children
And let's not forget:
Dumb parents => worse nutrition => shorter children
+
Dumb parents => dumb children => poor children
/A
What will they do after they've finished with Y and Z, they'll have no more letters of the alphabet then. It'll be a disaster
Then Norway will rise to world domination with the AE Window System.
The first thing that will then happen is that Slashcode will be eradicated; because that piece of Perl junk won't let me use a proper Æ in comments.
/A
The Danish name for a fruit machine translates into a one-armed thief. Now show me a slot machine that doesn't steal the gamblers money, that would be news.
/A
So... in this macho atmosphere, reeking of testosterone... the punishment for not being that the bug meet is that... YOU DON'T HAVE TO FIX YOUR BUGS UNTIL THE NEXT MAJOR RELEASE?????????
It seems the MS developers consider it a punishment to ship a product with a bug in the parts they are responsible for, and you consider it a reward. What does that say about the MS developers, and what does that say about you?
... which prefers to remain anonymous.
/A
You *know* it's technology for technology's sake and a complete waste of your time when the webpage needs a "brightness adjuster".
/A
This is patents doing what they do best: Preventing technology from being used!
manyoso: Hmmm. Seems to me that this 'art' has been around since the beginning of Unix. Hell, Microsoft has been providing a form of this 'art' with NT and 2000 for quite sometime. It's called permissions!
The difference being that you will not have root rights on your own hardware.
To use HTML, a mobile device would have to limit itself to some arbitrary subset, which is bad because every device manufacturer would choose a different subset. To standardize on an XHTML subset actually sounds like a good idea.
I'm sure those web browsers are great as long as you stick to the particular HTML subset that they can handle well. And I'm sure they suck when you don't.
Ouch! They must have worked very hard at Redmond to make it *that* bad, given that the obvious string concat implementation is O(n). That would make naive concatentation of N strings K characters each O(K N^3).
But that is a language implementation problem. Not a language problem per se. Nothing's to stop the KBasic team from doing better. They could do a lot better, actually: with a little thought and a little exponential overallocation in the string implementation, naive concatenation of N K-character strings would be only O(K N lg(K N)).
/A
Hmmm, and where can I get this wonderful managed security? Why look, Bruce himself sells it! What a happy surprise ...
Of course. He
Ok, so he could have chosen to do step 2 and 3 in reverse order. (Actually, maybe he did and we just didn't notice?) But that doesn't invalidate the analysis. Nor the product.
/A
>holiday trips to Mars
If you're lucky to have a holiday that lasts several years, that is.
Hit the throttle, accelerate at 10 m/s2 the first day, decelerate at the same rate the second day, and you're there, and you even had gravity all the way. Why would you want to spend years on Mars? :-)
Sure, with today's staged rockets and slingshot technology the voyage alone would take years. Which is exactly why I wouldn't go today.
/A
What do you mean by "more efficient" ways of space travel? Let's investigate your ideas one by one :
Whoops. My title "Build the space infrastructure first" was slightly off-key. It should have said "Invent the space infrastructure first". It wasn't that I have a lot of brilliant ideas as to how space travel should really be done, and every single one of the specific ideas that I hintet at may well be useless. But if these ideas don't pan out then we need to get some other great idea, because today we really don't have a clue on how to do space travel in a reasonably efficient manner.
The fundamental problem of how to get materials, mainly fuel, free of earth's gravitational pull, will not be solved by going to Pluto. Such a project would just use the same old big multi-stage rocket, which is proven technology. It is simpler to buy a huge fuel tank and fill it up than it is to develop some future technology that we haven't even begun to imagine yet. But we need the latter if it's ever going to be more than the odd one-off giga-dollar expidition.
Repair Facilities : In space, you don't "repair" something : you replace them. It's WAAY cheaper to be redundant in components than to build elaborate facilities to repair things..
Quite right. But again, it doesn't scale. So we'll just have to figure out how to build simple, generic repair facilities instead of elaborate, bulky, a-tool-for-every-job facilities.
A better challenge than going to Pluto would be this: Build a self-sustaining space station. In orbit, on the Moon, on Mars, I don't care. Any which way it is a formidable challenge, but in return we would learn a lot, not just about space, but also about ourselves, ecosystems, life.
/A
Space travel is so incredibly expensive.
Rather than spending the big $$$ on going into space now, they should be spent on finding more efficient means of space travel.
They should be researching propulsion systems, building waystations in orbit, building lunar refueling and repair facilities, gas^H^H^Hwater stations for fusion reactors etc.
In 100 years, when we have a reasonable handle on these things, we can begin sending probes into deep space, taking holiday trips to Mars and much much more, and we will learn infinitely more than we would learn from a Pluto mission today. To attempt these things before we have developed decent space travel logistics, is just a waste of resources.
OK, so you and I will not live to see the the results of such a long term plan. But then space exploration is inherently long term.
/A
"Enter a check", huh?
The chess 'bot would wake up and tell you
Seriously, ambiguity is a big problem with natural-language interfaces. How is the computer to know if you wanted to do something involving your checking account or if you want to add an integrity constraint to a computer program you are writing? Or if you are playing a chess game and trying to make an illegal move ...
Natural language comprehension is an AI-complete problem. That doesn't mean useful approximations can't be done. But they only work if you keep commands simple and adhere to a computer-friendly style of expression. With requests of any complexety the risk of a misunderstanding is too great to trust interface to do anything that can't easily be undone. In this sense it is similar to DWIM interfaces (follow the link for a good anecdote).
No doubt natural language interfaces will find its niches, and someday many people may even be using natural language interfaces exclusively. But when these people need to do something more complex or risky, they will need to turn to a hacker who masters some other arcane but concise command language.
/A
There is a conflict of interests at work here:
Full disclosure is suboptimal because people have better things to do with their time than upgrade and patch software. No disclosure is suboptimal because there is no pressure on software vendors to fix holes.
Full disclosure works well with stable software. Eventually bugs get fixed and the continuous public review will have created a secure product which can be used for years and years.
It is with rapidly changing software there is a problem. And in these days where "internet time" is a valid excuse for anything, rapidly changing software is what we have lots of. For this, we need a good idea for how to strike a balance between the two extremes, full disclosure and full secrecy.
One idea is to have a "waiting period". When someones discovers a security problem they inform the vendor but wait some about of time before informing the public, say 1 month. That way not only will a fix be ready place when the problem is publicized, but frequent upgraders may already have the patched software running. The software vendor, knowing that the problem will be publicized at the end of the waiting period, still have an incentive to get it fixed.
Of course this doesn't help the masses with years-old software. Someone please get an even better idea!
/A