Apple's.Mac is mostly a collection of open standards.
Brilliant, dude: so you can invest a few man-years to develop software that Apple can choose to break with their next update and that then provides unsupported access to a $100 service that other companies largely provide for free. This amazing degree of openness on the part of Apple is hard to beat!
They're obviously about ready to take that mobile.
They already did: if you buy a Mac and an iPhone, then you can access your.Mac data from your iPhone. Will the wonders never cease!
Apple hasn't done a lot with it beyond those things to date, but hints that is about to change... I'd say they have a head-start on Microsoft, yet again.
They sure do have a head start on Microsoft, including the "it will only work well with our own OS" part.
I think the real leaders in this area are the companies that have figured out how to offer these services in an OS-neutral way and how to integrate mobile and desktop usage. Neither Microsoft nor Apple have done that.
You know, there are people worth spending time on exposing as fraudsters (vice presidents, CEOs of software companies, etc.), and there are people who aren't (spoon benders, fortune tellers, etc.).
James Randi's time would be better spent on the former group of people.
You have it backwards. The copyright holder can terminate a license anytime they wish and they are not bound by the GPL.
You may mean the right thing, but you are expressing it incorrectly. Apple cannot retroactively "terminate" the license; the GPL is irrevocable. What they can do is stop releasing new versions under the GPL and instead release new versions (or old versions) under a different license.
It all seems so straight forward if you accept, as you say you do, the relationship between compression and optimal prediction
Your mistake is that you're equating optimal prediction with intelligence. Many natural and artificial systems learn to make optimal predictions, yet aren't intelligent. And intelligent systems often don't make optimal predictions.
I don't understand what your issue with the Hutter Prize is.
It is precisely what I said: it misleads people into equating optimal prediction with intelligence.
Exposing Uri Geller's spoon bending as fake is like exposing Pamela Anderson's breasts as fake. What's the point? It spoils the fun, and those who still think it's real aren't going to be convinced anyway.
Then you get to watch the spinning beach ball until (1) you get an ex girlfriend (note: for that, you first need a current one) or (2) the battery runs out (with the latter being far more likely).
Twice the bandwidth at 1/10 the price and cost. Seems like a good deal to me.
(Remember: this is the same AT&T that was trying to tell us that packet switching networks would never work for voice and video and that we should all be investing enormous amounts of money in expensive circuit switching technology. Now they are trying to shoehorn their hare-brained ideas into QoS guarantees on the Internet.)
Clearly being able to predict the next symbol in a stream coming from a corpus of human knowledge is relevant to AI and such prediction is the core technique of compression.
No, it is not "clearly" relevant at all.
If you have an unknown environment the fact that it is unknown means you are making educated guesses as to maximum likelihoods, or "expectation maximization".
You're mixing up a bunch of things. First, maximum likelihood and expectation maximization are different concepts. Second, maximum likelihood decisions are not optimal in general. Third, in actual experiments, humans often don't make either optimal or maximum likelihood decisions, meaning that it's odd to use anything related to them as a measure of intelligence.
Are you saying NLP model perplexity has never proven to be very useful in evaluating AI systems?
Power consumption and price are also very useful in evaluating AI systems, they just have nothing to do with AI. Perplexity is a measure of how good a statistical natural language model matches word frequencies, not of AI.
He proves that the optimal behavior of an agent (an interactive system that receives a reward signal from an unknown environment) is to guess that the environement is most likely computed by the shortest possible program that is consistent with the behavior observed so far.
But the task of compressing Wikipedia character by character is thoroughly irrelevant to human intelligence. Humans are bad at it, so it's not a characteristic of human intelligence, and if you had a very high quality compressor, even if it used some kind of internal model of the meaning of the text, you still couldn't use it as an AI system.
Another problem with the whole approach is the assumption of optimality; in fact, intelligent behavior is unlikely to be optimal in any given environment.
Hutter proved that the optimal behavior of a goal seeking agent in an unknown but computable environment is to guess at each step that the environment is controlled by the shortest program consistent with all interaction so far.
Well, one should keep in mind that the connections between AI and compression have been known far longer, but have never turned out to be particularly useful for building AI systems. Hutter's point is merely a variant of these previous theories, and there is no reason to believe that it will lead to AI any more than previous attempts.
You just pile the unsupported assumptions on top of each other. How do you know that your sample is biased towards the better applicants?
Because there are objective ways of measuring and quantifying that, and companies do that because it's an important part of hiring.
You can't reliably improve a process if you have no objective way to measure the outcome.
That's why companies objectively measure the outcome. In fact, that's the primary function of middle management. Furthermore, it's not just each company that does this, psychologists and social scientists go into companies, perform detailed observations, and generalize and publish the results to help everybody improve their processes.
The issue isn't whether I (or anyone else) can come up with a better process, it's whether the process you use is good enough to allow you to draw conclusions about the qualifications of people you have never met and know nothing about. Nothing you've said has convinced me of that.
Well, then you'll have to remain unconvinced. That doesn't change the fact that the majority of applicants for programming positions is ill-prepared. In fact, they often don't even make it to the subjective issues like teamwork and performance under pressure, they trip up on simple, objective questions related to Java, C++, or software engineering.
If you look at the description of PAQ, you'll see that it doesn't attempt to understand the text; it's just a grab-bag of other compression techniques mixed together. While that is nice for compression, it doesn't really advance the state of the art in AI.
If you want to draw a more general conclusion, you need a more general sample to be valid.
No, I don't need a "more general sample". The sample I have seen has been biased towards the better applicants, therefore, the rest of the industry gets worse applicants.
It sounds like you believe your subjective process is a good one
Where did I say that the process was "good"? Actually, it sucks: it's a lot of work, and it has a high error rate. But it's the best anybody has ever been able to come up with.
and aren't really interested in finding out if it's flawed or could be improved.
Hiring is so important to companies that they are investing a huge amount of time and money in trying to improve the process. If you can come up with a demonstrably better process, let the world know about it.
That pretty much confirms there is no objective criteria.
There are plenty of objective criteria: you need to have the right degrees, experience, test scores, and recommendations. Only people meeting those objective criteria make it to interviews. And at that point, there is a final selection based on subjective criteria. You don't seriously expect that companies hire people without a judgment call by interviewers?
Do you select your friends based only on objective criteria? Well, hiring is no different.
And what's a "PDA Phone"? This would be a good place for one of those "The 90s called and it wants its X back" jokes, but how is the 90s supposed to call if you've got its goddamn PDA phone.
What makes a phone a "PDA phone" is its requirement to sync with the desktop; that's exactly what makes the iPhone an anachronism. This is the 21st century, and these days we sync OTA using 3G networks, or we simply assume "always on" connectivity.
Back to the proposed wager. I'll bet you my foolish optimism against your lack of imagination.
Actually, it's you who lacks imagination: except for its looks and UI, the iPhone is a throwback to the times of the Palm: a bunch of apps with desktop synchronization.
Look, I have held and tried; the Neo is actually small and light enough to keep in your pocket, these other devices simply aren't. They are also a lot more expensive, if you can get them at all.
For me, the Neo hits a sweet spot in terms of size, price, and screen quality that no prior phone has been able to achieve.
When I think "high quality toilets", the places that come to mind are the US (Stevens Institute of Technology, birthplace of the low flush toilet), Japan (high tech computerized toilets), and Germany (all-round well-engineered bathroom technology). The Italians and French maybe have a historical claim on toilet technology. But there's no way I'd want a piece of Russian-built toilet technology anywhere near my private parts.
Slashdot Mirror
The really funny thing is, that the one valid complaint you could level against .Mac you got wrong!
.Mac, and you synchronize your iPhone against your Mac.
.Mac is expensive and it is a Mac-centric service.
No, I didn't. You synchronize your Mac against
And my other complaints are quite valid:
Apple's .Mac is mostly a collection of open standards.
.Mac data from your iPhone. Will the wonders never cease!
Brilliant, dude: so you can invest a few man-years to develop software that Apple can choose to break with their next update and that then provides unsupported access to a $100 service that other companies largely provide for free. This amazing degree of openness on the part of Apple is hard to beat!
They're obviously about ready to take that mobile.
They already did: if you buy a Mac and an iPhone, then you can access your
Apple hasn't done a lot with it beyond those things to date, but hints that is about to change... I'd say they have a head-start on Microsoft, yet again.
They sure do have a head start on Microsoft, including the "it will only work well with our own OS" part.
I think the real leaders in this area are the companies that have figured out how to offer these services in an OS-neutral way and how to integrate mobile and desktop usage. Neither Microsoft nor Apple have done that.
You know, there are people worth spending time on exposing as fraudsters (vice presidents, CEOs of software companies, etc.), and there are people who aren't (spoon benders, fortune tellers, etc.).
James Randi's time would be better spent on the former group of people.
Monkey see, monkey do, monkey dance.
You have it backwards. The copyright holder can terminate a license anytime they wish and they are not bound by the GPL.
You may mean the right thing, but you are expressing it incorrectly. Apple cannot retroactively "terminate" the license; the GPL is irrevocable. What they can do is stop releasing new versions under the GPL and instead release new versions (or old versions) under a different license.
If Apple were to put a GPL3 CUPS on the AppleTV or osme similar product they'd probably waive some patent rights by selling the product.
The GPLv3 applies only to the software that is licensed under it, so Apple doesn't lose patent rights to anything other than CUPS-related patents.
Note that the Apache 2 license also has patent clauses; I don't see Apple being terribly concerned about that.
It all seems so straight forward if you accept, as you say you do, the relationship between compression and optimal prediction
Your mistake is that you're equating optimal prediction with intelligence. Many natural and artificial systems learn to make optimal predictions, yet aren't intelligent. And intelligent systems often don't make optimal predictions.
I don't understand what your issue with the Hutter Prize is.
It is precisely what I said: it misleads people into equating optimal prediction with intelligence.
Exposing Uri Geller's spoon bending as fake is like exposing Pamela Anderson's breasts as fake. What's the point? It spoils the fun, and those who still think it's real aren't going to be convinced anyway.
If you want a Nano-style phone--slim, metal case, good music player--get a Sony W880i. Bonus: it runs Java apps and it supports 3G.
Then you get to watch the spinning beach ball until (1) you get an ex girlfriend (note: for that, you first need a current one) or (2) the battery runs out (with the latter being far more likely).
Twice the bandwidth at 1/10 the price and cost. Seems like a good deal to me.
(Remember: this is the same AT&T that was trying to tell us that packet switching networks would never work for voice and video and that we should all be investing enormous amounts of money in expensive circuit switching technology. Now they are trying to shoehorn their hare-brained ideas into QoS guarantees on the Internet.)
Clearly being able to predict the next symbol in a stream coming from a corpus of human knowledge is relevant to AI and such prediction is the core technique of compression.
No, it is not "clearly" relevant at all.
If you have an unknown environment the fact that it is unknown means you are making educated guesses as to maximum likelihoods, or "expectation maximization".
You're mixing up a bunch of things. First, maximum likelihood and expectation maximization are different concepts. Second, maximum likelihood decisions are not optimal in general. Third, in actual experiments, humans often don't make either optimal or maximum likelihood decisions, meaning that it's odd to use anything related to them as a measure of intelligence.
Are you saying NLP model perplexity has never proven to be very useful in evaluating AI systems?
Power consumption and price are also very useful in evaluating AI systems, they just have nothing to do with AI. Perplexity is a measure of how good a statistical natural language model matches word frequencies, not of AI.
The value of the Hutter Prize is in the use of compression ratio to provide a figure of merit for AI that is very good.
The same figure of merit has existed for decades before, and it has never proven to be very useful in evaluating AI systems.
That value has been proven in various ways -- mathematically and practically.
Really? Various practical ways? Like what? Where has a Hutter prize related advance been demonstrably linked to an advance in NLP?
He proves that the optimal behavior of an agent (an interactive system that receives a reward signal from an unknown environment) is to guess that the environement is most likely computed by the shortest possible program that is consistent with the behavior observed so far.
But the task of compressing Wikipedia character by character is thoroughly irrelevant to human intelligence. Humans are bad at it, so it's not a characteristic of human intelligence, and if you had a very high quality compressor, even if it used some kind of internal model of the meaning of the text, you still couldn't use it as an AI system.
Another problem with the whole approach is the assumption of optimality; in fact, intelligent behavior is unlikely to be optimal in any given environment.
Hutter proved that the optimal behavior of a goal seeking agent in an unknown but computable environment is to guess at each step that the environment is controlled by the shortest program consistent with all interaction so far.
Well, one should keep in mind that the connections between AI and compression have been known far longer, but have never turned out to be particularly useful for building AI systems. Hutter's point is merely a variant of these previous theories, and there is no reason to believe that it will lead to AI any more than previous attempts.
You just pile the unsupported assumptions on top of each other. How do you know that your sample is biased towards the better applicants?
Because there are objective ways of measuring and quantifying that, and companies do that because it's an important part of hiring.
You can't reliably improve a process if you have no objective way to measure the outcome.
That's why companies objectively measure the outcome. In fact, that's the primary function of middle management. Furthermore, it's not just each company that does this, psychologists and social scientists go into companies, perform detailed observations, and generalize and publish the results to help everybody improve their processes.
The issue isn't whether I (or anyone else) can come up with a better process, it's whether the process you use is good enough to allow you to draw conclusions about the qualifications of people you have never met and know nothing about. Nothing you've said has convinced me of that.
Well, then you'll have to remain unconvinced. That doesn't change the fact that the majority of applicants for programming positions is ill-prepared. In fact, they often don't even make it to the subjective issues like teamwork and performance under pressure, they trip up on simple, objective questions related to Java, C++, or software engineering.
If you look at the description of PAQ, you'll see that it doesn't attempt to understand the text; it's just a grab-bag of other compression techniques mixed together. While that is nice for compression, it doesn't really advance the state of the art in AI.
If you want to draw a more general conclusion, you need a more general sample to be valid.
No, I don't need a "more general sample". The sample I have seen has been biased towards the better applicants, therefore, the rest of the industry gets worse applicants.
It sounds like you believe your subjective process is a good one
Where did I say that the process was "good"? Actually, it sucks: it's a lot of work, and it has a high error rate. But it's the best anybody has ever been able to come up with.
and aren't really interested in finding out if it's flawed or could be improved.
Hiring is so important to companies that they are investing a huge amount of time and money in trying to improve the process. If you can come up with a demonstrably better process, let the world know about it.
That pretty much confirms there is no objective criteria.
There are plenty of objective criteria: you need to have the right degrees, experience, test scores, and recommendations. Only people meeting those objective criteria make it to interviews. And at that point, there is a final selection based on subjective criteria. You don't seriously expect that companies hire people without a judgment call by interviewers?
Do you select your friends based only on objective criteria? Well, hiring is no different.
How can Apple win through monopolistic practices when they didn't have any market share in smartphones AT ALL a little over a week ago?
Apple is leveraging their near-monopoly in the music player market to get into the phone market.
Indeed. Kudos to Apple for showing the rest of the market how it is done.
Where exactly do you think Apple innovated? The iPhone seems little more than a Treo with a touch screen and an updated UI.
And what's a "PDA Phone"? This would be a good place for one of those "The 90s called and it wants its X back" jokes, but how is the 90s supposed to call if you've got its goddamn PDA phone.
What makes a phone a "PDA phone" is its requirement to sync with the desktop; that's exactly what makes the iPhone an anachronism. This is the 21st century, and these days we sync OTA using 3G networks, or we simply assume "always on" connectivity.
Back to the proposed wager. I'll bet you my foolish optimism against your lack of imagination.
Actually, it's you who lacks imagination: except for its looks and UI, the iPhone is a throwback to the times of the Palm: a bunch of apps with desktop synchronization.
Look, I have held and tried; the Neo is actually small and light enough to keep in your pocket, these other devices simply aren't. They are also a lot more expensive, if you can get them at all.
For me, the Neo hits a sweet spot in terms of size, price, and screen quality that no prior phone has been able to achieve.
Well, the iPhone is definitely not giving me the apps I want, while phones costing a small fraction of the iPhone do.
When I think "high quality toilets", the places that come to mind are the US (Stevens Institute of Technology, birthplace of the low flush toilet), Japan (high tech computerized toilets), and Germany (all-round well-engineered bathroom technology). The Italians and French maybe have a historical claim on toilet technology. But there's no way I'd want a piece of Russian-built toilet technology anywhere near my private parts.