One thing to remember is that there may be times when you want to capture more than a human eye can take in all at once.
Another thing to remember is that the eye throws away the vast majority of what it sees before it ever hits your brain.
I was at the San Francisco Exploratorium last week, and saw a nifty demonstration of this. They have a projector displaying a scene of a street front. Every few seconds, it changes something in the image. The changes are actually quite dramatic (removing or adding a parking meter, replacing a lamp post with a colorful sign, etc.) If the program blacked out the screen for a split second before each change, you'd see the change every time. If it didn't, you'd be lucky to catch the change one time in five.
So if you take the amount of information the eye is actually sending to your brain, and rendered that many pixels of a scene, it would look extremely jagged. The camera would have to capture way more information than your eye is taking in, in order to do the entire scene, because your eye darts around a lot.
Let's not bicker and argue over who hypocrited who. I was being facetious, he's not a hypocrite, and Craigslist really is the coolest thing on Web 1.0.
It wouldn't be hard to let a user do something to click an "enable ads" button. Each ad could have a link near it saying, "Okay, that's enough support for one day." Since users would have to take proactive measures to enable ads, and could easily disable them again if they decided they were annoying, I don't think my desire to see ads would conflicat at all with your desire to not see ads.
I'd go so far as to suggest that Wikipedia adopt the same system. I'd enable ads for them, too.
But the guy is a total hypocrite! He talks about giving users what they want, and as a user, I want ads on Craigslist! I like them, and I'd like to know that they're earning money from my traffic, rather than being drained by it. So where is the little, unobtrusive button at the bottom of the page that says, "support Craigslist by enabling ads?" I'm looking for it, and I don't see it.
If they wanted to do this, and give most of the money to charity, that would be even cooler.
Hmmm... so, here's an anti-business model somebody could try. Create the "Progressive Advertising Network", which would be strictly opt-in by users. People who wanted to see the ads would click an enable button, and they'd show up on every page that had them. The ads would be a mixture of ads for non-profit groups and standard ads which subsidized the non-profit ads. Excess profit would be donated to some reasonably noble cause.
In response to your conclusion: Big business + the ability to pass legislation benefitting yourself at the expense of the general public does equal corruption.
Your claim that incumbency is merely a matter of name recognition is demonstrably false. Incumbents also get the vast majority of campaign contributions (especially from business). Incumbents also have the free publicity that comes from being a member of Congress. Incumbents have a great deal more sway with the people in their state who draw up district boundaries. The first one is the most important, because the candidate who spends the most usually gets the most votes.
Nor do your arguments against bankruptcy laws make much sense. "Do you really think it's 'fair' that people be allowed to absolve their own irresponsibility simply by declaring bankruptcy?" Frankly, yes.
1) Credit card companies frequently make stupid loans to high-risk individuals, with the intention of letting them run up debts that will be very difficult to repay. 2) Most people who file for bankruptcy aren't doomed by their "personal irresponsibility," but by a sudden change in their life situation, such as divorce or unexpected medical bills. Oh, wait, divorce is de facto proof of personal irresponsibility. Go ahead and say it. I'm sure you're thinking it. 3) At the time the law was passed, the lending companies that benefitted were already posting record profits.
A corporation is not a person, and shouldn't have the same powers to petition the government that normal people do. If you can't see the problems that the alternative causes, you're trying very, very hard to avoid looking for them.
Corruption is very much a part of the American system. It's not so egalitarian as the corruption in most countries, where local officials and low-level functionaries get in on it. But in the highest levels of government, if you're a corporation, you can come to a senator with a bundle of "contributions from our supportive workers", and then come to him again with a piece of legislation that conveys the benefit of a government granted monopoly, or a government contract.
Not only is this sort of thing common and destructive to our political process, it's often the best investment an industry can make. Before September 11th, the airline industry made millions in political contributions to influential senators. After September 11th, they got their payout: Billions of dollars to "save the airlines." Some of the money went to beefing up airport security (which, in retrospect, would have been a good investment, and hence the free market should have automagically taken care of it). But most of it went to bail out investors. Little of it went to saving jobs, and the airlines cut tens of thousands of jobs in the name of "efficiency".
More recently, we have a bankruptcy reform bill pretty much designed from the ground up by the credit card industry, an energy bill designed to funnel billions of dollars to the energy industry in exchange for basically doing what they've always been doing, and a "reform" of the Endangered Species Act designed solely to gut the intent of the act and promote the interests of land developers.
And guess what? It is inherent in the system, not the people. If we stepped in today and forced public financing of elections, corporations couldn't use campaign contributions as leverage to get their way. Period. It would also lower the current un-democratic incumbency rates, making long-term investments in individual politicians much more risky. Or would you like to try and argue that our 98% re-election rate just shows that our electorate always chooses the best person for the job the first time?
Yes, there is such a thing as human nature. It has good in it, and it has evil in it. Systems can be designed to promote the best in human nature, or they can be designed to subvert good intentions in favor of raw, consuming greed. We're currently living in the latter sort of system, and until we fix that system, we can try electing better people, and they'll get chewed up and spit out by it.
But what if (just toy with the idea for a moment or two) Vista isn't the most secure OS ever? Call me crazy, but I think there may be a tiny possibility that Microsoft is saying positive things about its products that may not be true.
C'mon, what are they going to say? "Vista is basically less secure than XP was, but we try to compensate for that by popping up lots and lots of messages." People wouldn't exactly be lining up to buy that, would they?
It's asinine to address Bill directly. I mean, Bill Gates kicking back and relaxing by reading Slashdot? Sure, he's a geek, but I still don't see it. Also, in the unlikely event that he is reading, what were the odds that he would let random comments from the peanut gallery "deter him for one second?" Your accolades, like the catcalls of other posters, are easy enough to brush off.
I don't think Bill Gates will be looked back upon as a "great man". While he was certainly an important force shaping our world, he got that influence by a combination of business skill, ruthlessness, and dumb luck. His net effect on technology might have been positive, or it might have led the whole system down a blind alley filled with limited choices and badly interoperating proprietary technologies.
He'll always have his supporters and his detractors, and his historical legacy will depend on whose interests the history books happen to be serving at the moment. Since they often serve wealth, I imagine that overall he will be treated well.
1) That's about 1/6th of his net worth. In some ways, it's a less impressive gesture than if somebody of far lesser means gave away 1/6th of his net worth. After all, how many people could give away 98% of their wealth and still be in the billionaire's club?
2) It's a long-running debate whether such vast collections of wealth can ever be truly "earned".
3) It's a long-running argument whether Microsoft got its money in ways that promoted the overall technology ecosystem or hindered it.
4) The social inequities these vast accumulations of wealth represent are the root of many of the ills in our society.
I don't think this will help with "rapid prototyping," since it has to add an additional compilation step to your change, test, change cycle.
Nor do I think most projects will be able to depend entirely upon it. Looking at the limitations listed in the announcement, it seems like it takes away many of the things that make Python engaging as a language. What seems to remain? Imagine taking a compiling C program, translating it (as literally as possible) into python, and then let this program translate it into machine code for you.
I'm probably overstating the case. There will be certain subsets of code that might benefit from this (where people need the performance, and are willing to give up architecture independence to get it). But even if this thing matures greatly (and I'd like to see that happen), I don't think it will become ubiquitous.
Please step back and try to see the point I've been trying to make this whole discussion: If you don't know the algorithm that will get you the right results (AI), or your source data isn't fine-grained and accurate enough (weather simulation), then all the speedup in the world is only going to get you the wrong answers faster.
This is true even for neural networks and genetic algorithms. For any given neural net configuration and set of inputs, you can only train it so much before your results stop improving (and eventually start getting worse due to overfitting). If your NN isn't properly configured (not enough layers, or an inappropriate number of nodes for each layer) or if you've chosen the wrong features as input, or if your training data is warped, the NN won't get good results. All faster hardware can do is get you to "the best you can do with what you have" more quickly.
Of course, faster processors let you do things like use genetic algorithms to select the ideal configuration for the neural net. But even genetic algorithms aren't magic. They're just one possible shortcut for exploring a large number of hypotheses. At best, GA can only match brute force of all enumerated hypotheses when it comes to the final performance of the selected algorithm. You still have to design your hypothesis space in such a way that a good solution lies within that space.
The problems I was referring to strike me as the sort where we haven't developed ideal approaches to the problem, and running the current approaches on beefier hardware aren't going to close the gap. Hence, my indignant response. My hunch is that, once we do find the proper algorithms for them, many of them will run surprisingly well on current hardware.
Fair enough. But C does provide about as little abstraction as one can expect from a language. I would argue (with some expectation of objections) that the greater the amount of abstraction, the more functionality a programmer can readily keep in his head. Say you have a subroutine that you can do the simple, slow, easily understood way, or do it the blazing-fast, well optimized, difficult to comprehend way. Which one you go for depends on several things, one of which is, "how likely is it that this code will need to be modified in the future?" If it's nice and modular, and you can wall it off so nobody need touch it again (and the performance boost is worth it) then go ahead, and shut the door behind you.
On the other hand, if it will require future programmers to understand and modify it, and you still need the performance boost, then the additional inherent complexity the programmers will have to memorize necessarily comes at the expense of something else.
Therefore, I believe that a simple, easily-grokked codebase is important. I think interpreted languages (esp. my favorites, Python and Ruby) allow for such simplicity. It might not be an overriding factor, depending on the job. But it's something to account for.
Exactly. The advantage of better hardware is the ability to either do bigger jobs, or do them in a more effective way (using better algorithms, for example). The problems described by the original poster are, for the most part, resistant to brute force (at least on the hardware we can reasonably expect in my lifetime). In many of the cases he proposes, it's not even certain what algorithm we'd throw at it if somebody dropped a Pentium 10 300-PetaHertz machine in our laps.
In your case, the key words are "real time". You knew exactly what algorithms would do the job, and you just needed a relatively trivial increase in speed to make it happen.
No. What you describe is an entirely different piece of software. Instead of running dog_brain.exe, you're running whole_world_to_find_ideal_dog_brain_via_genetic_al gorithm.exe. I'm talking about taking the same algorithm, and simply running it on faster hardware. If your text parsing system relies on the Viterbi Algorithm to classify words, then faster hardware means only that you can tackle bigger projects with the same level of accuracy. It doesn't magically turn your program into a perfect parser.
Okay, a more obvious example. Take the TI-85, the calculator I used in high school. Speed it up by a factor of a million. It doesn't get any more accurate, because it's still using the same number of bits to calculate. It doesn't magically gain the ability to solve problems that it couldn't before (unless the problem are time or memory-bound). The advantage to hardware improvements come from the ability to use the right algorithm, instead of hacks to bring in an answer in an acceptable amount of time.
Okay, I should have thrown out a few caveats about algorithmic complexity. Pure lookahead is indeed ideal, given sufficient time to explore *all* options. But pure lookahead to a depth of n? We don't know how deep a pure lookahead algorithm would have to go to beat, say, Kasparov. Might be a new way to rate chess players.
No human is going to wait around (or even live) long enough for his computer opponent to completely map out the entire chess game space, and given the branching involved, speeding up by a factor of a thousand only gives you a few more looks ahead. It's safe to assume that hardware will never be fast enough to fully map out chess space.
My point is only that, once hardware reaches a certain speed, its speed ceases to be the primary constraint, and we're left with the best performance a given algorithm can give us. You seemed to be under the impression that the solution to any AI problem was to throw more hardware at it.
I was right with you until you started gushing about the beauty of assembly. You might be able to write a single, heavily-used subroutine in assembler, and make a thing of beauty out of it. But the moment you try to do something truly ambitious, you abso-friggin-loutely need layers of abstraction. CPU cycles are one thing you don't have an unlimited amount of. Brain cycles are another.
It wasn't all that absurd. I was just saying that *if* the tax brackets cause the sort of massive disincentives that neocons claim, *then* it's absurd to put the strongest single change in incentives that close to the average person's earning power. I fully agree, nobody tries to make less in order to stay under the cap.
You seem to be under the impression that these problems you cite display inadequacies in the hardware, rather than the software. But, in the words of some fictional professor from a book I can't remember: "If you speed up a dog's brain by a factor of a million, you'll have a machine that takes only three nanoseconds to decide to sniff your crotch." Given the current software and algorithms available, more computing power alone wouldn't solve any of the problems you describe.
I really do try to be more careful in my language than that. I meant to say that there is little motivation for the departing employee to train his replacement well.
If you're a balloon company, and your most difficult choices are what color balloons to make and who to get to ship them, then sure, rule by committee. Anything more complicated needs leadership.
I'm going to remember this conversation next time I hear some neocon ideologue yammering about how liberals are a bunch of paternalists who don't trust people to make the right decisions.
Sorry, snarky Marxist Bryce woke up today. But it does strike me as rather paternalistic, the idea that we have a few bold titans running around, and that the value of their hard work and bold decisions somehow justify a salary five hundred times that of their average employee.
I've seen individuals make good decisions, and I've seen individuals make bad decisions. I've seen committee work where a great idea was dumbed down to worthlessness, and I've seen the same process defang or eliminate really bad ideas, and substantially improve good ones. On average, I think committees perform very well, so long as their constituents are motivated by an actual interest in the subject matter and decision-making process, rather than a desire to sound creative, or just to advance their careers by being on a committee. How people were chosen for what committee work--and how they are rewarded for that work--is something that the employees would come up with at the beginning, and would be subject to tweaking.
Perhaps in some cases, people would prefer a committee of one. One person would be empowered to make certain decisions. In that case, the only difference between their operation and that of a normal company would be the fact that under worker rule, the decision maker could be recalled.
Your "bad committee decision" anecdote is certainly important. But it sounds like a situation where everyone had incentive to learn just enough to cast an "informed vote", with nobody having the extra incentive to dig deeply into the issue and present the findings to the voters. So while the bad decision may stem from the details of your process, and may not apply to all committee-style decision making.
Your assertion that we are designed to break ourselves up into leaders and followers is incorrect, or at least requires some huge caveats. You can look at baboon societies and say, "Yeah, you can always find a strict heirarchy," and draw your conclusions from that. Or you can look at the studies showing that baboons lower down the pecking order have significantly higher stress, worse health, and shorter lifespans. I honestly believe that, when we decide that a few leaders get all the power, and the rest of us simply follow their decisions, that we subject the vast majority of human beings to the same effects: higher stress, worse health, and shorter lifespans. Most any study on the subject of wealth and health will tell you that poverty-level wages are little more than a slow death sentence. But they also show that many far poorer--but far more egalitarian--countries than ours have better health outcomes than we do.
Last point, then I have to run: You ask, "If everyone is busy boning up on decisions, then who will be doing the work?" I think you have it 360 degrees backwards. It's like criticizing a company for spending too much energy on employee training. If employees have a deeper understanding of how the overall structure fits together, their work will be more effective. The incentive in worker-owned companies would be for knowledge to be spread broadly, because everyone has a stake in making the company better. Of course, there are also incentives for hoarding, which is why a culture of transparency would be critical for success.
I would argue that the companies doing the outsourcing aren't bearing the full economic costs of the transition, and that there are other costs that need to be factored in.
First, there is the cost to the employees of finding a new job, which is never fully reflected in severance pay. Severance pay probably only covers the amount of time the employee will be idle. Other costs include retraining the employee to fit the new job (paid by his new employer), the psychological effects of having his job terminated (paid by him, his family, and society at large), the costs of moving if the new job requires it, and the loss of community if he does have to move. This last item is important: while a loss of community has all sorts of ill effects, the expectation of losing it several more times over the course of your work history is even worse. Ask anyone raised by parents who got moved around a lot by the military.
People who expect to move away end up less trusting, less willing to invest in new relationships, less willing to participate in community functions or worry about local issues. There was a time when the rich still had some sort of shared community life with their workers, and it was a time when employers had loyalty to their employees. But now those who own the wealth--and decide how it will be used--are more distanced from the rest of us, and have no stake in that shared community life. So they'll happily move jobs to Outer Ohfuckitstan to earn an extra 1% return on their investment, without regard for the problems these moves cause the wider society.
The country that benefits from the new jobs also receives increased inequality between the well-off and the poor. India's rise has been accompanied by increased tension between wealthy urban areas and poor rural areas. Back when everyone was poor, the rural areas weren't likely to complain about not having electricity, because few did have it, and it's hard to want things you don't see anyone else enjoying. While an increase in the standard of living is wonderful, it is poisonous to a society when extreme inequalities in wealth exist. Of course, by this metric, our own society is very, very unhealthy.
There is another cost that employers do pay, but rarely fully account for when planning an offshoring: the loss of expertise. No matter how well the departing employees train their replacements (and what's their motivation to do a good job?) it will be years before the replacements can match the original expertise. Some of the knowledge lost in transition may never be rediscovered, leading to permanent inefficiencies.
No, I'm not a socialist. But I'm coming to the belief that collective ownership of wealth is a good thing. If Bank of America was owned by its employees, then their salaries would show up on the "income" side of the accounting ledger rather than the "expenses" side, and much of the motivation to outsource would evaporate. Otherwise, there is an inherent conflict of interest between management and labor: management wants to keep costs low, and labor is simply another expense to be minimized.
Additionally, a worker-owned setup is necessarily more efficient. Workers have the greatest incentive to make the company run as efficiently as possible, so management doesn't have to spend its time figuring out how to best motivate the employees. The workers would just tell them. Worker-owned companies are also more concerned with long-term stability, because there is no golden parachute waiting if they can just keep the stock prices from tanking in the next three quarters. There is only the promise of continued employment. Finally, when a few investors own a company, the goal is to maximize their profit. When workers own a company, they still want to maximize their returns, but they can accept a wider variety of payment: a cleaner environment, more free time to spend with their families, improvements in the social fabric of their communities, less stressful working conditions.
In short, collective ownership allows the wealth of a company to serve the many, not the few. So bring it on, free market boy.
One year's gross income? For a house? Fixer-upper or no, that's quite a find. Was it a foreclosure? Real estate market depressed in your area? Send Guido and Nunzio over to negotiate the sale?
Seriously, though. I admire the way you're living, and your ability to opt out of the overwork-to-overconsume treadmill. Most people end up getting into a house bigger than they can afford, and spend their whole lives busting ass to afford the house, so that they can say their houses are as big as the neighbor's houses. The neighbors, in turn, are spending their whole lives busting ass to pay off houses they can't afford either. But the pressure to "keep up with the Joneses" disappears once you stop seeing it as a competition.
Of course, this rat-race mentality leads to bigger and bigger houses, which makes the market unaffordable for everyone, even those who only want a place to live, rather than a status symbol. This glut of overbuilt housing isn't going away, and there are psychological and economic reasons for thinking that this housing bubble may never really "pop". So one option I never really see mentioned: splitting a house with another family. Most suburban houses are filled with crap of marginal utility, as well as copies of stuff their neighbors have. Combine households, and much of the stuff you own becomes redundant.
Other advantages: Shopping for groceries and cooking for eight doesn't take nearly twice the time of doing it for four. Only one lawn to mow. Kids have greater access to playmates and homework help. If both of you are two car families, there are probably ways to do without the fourth car. One Internet connection with twice the bandwidth is much cheaper than two separate connections. You can drop one of your cable subscriptions (but heck, why not drop both?).
Of course, there are downsides. The number of relationships to be managed goes up drastically. People may feel crowded in (though a lot of these new houses are positively huge). Splitting up the household could be messy if things go sour. But real estate has long since departed the realm of affordability for most people, especially in areas that don't require an hour or so of commuting to work. This is driving all sorts of non-American-Dream behavior, like the sudden rise in adults living with their parents. So I think that, as time goes on, more people will be considering shared living arrangements.
Slashdot Mirror
One thing to remember is that there may be times when you want to capture more than a human eye can take in all at once.
Another thing to remember is that the eye throws away the vast majority of what it sees before it ever hits your brain.
I was at the San Francisco Exploratorium last week, and saw a nifty demonstration of this. They have a projector displaying a scene of a street front. Every few seconds, it changes something in the image. The changes are actually quite dramatic (removing or adding a parking meter, replacing a lamp post with a colorful sign, etc.) If the program blacked out the screen for a split second before each change, you'd see the change every time. If it didn't, you'd be lucky to catch the change one time in five.
So if you take the amount of information the eye is actually sending to your brain, and rendered that many pixels of a scene, it would look extremely jagged. The camera would have to capture way more information than your eye is taking in, in order to do the entire scene, because your eye darts around a lot.
Let's not bicker and argue over who hypocrited who. I was being facetious, he's not a hypocrite, and Craigslist really is the coolest thing on Web 1.0.
It wouldn't be hard to let a user do something to click an "enable ads" button. Each ad could have a link near it saying, "Okay, that's enough support for one day." Since users would have to take proactive measures to enable ads, and could easily disable them again if they decided they were annoying, I don't think my desire to see ads would conflicat at all with your desire to not see ads.
I'd go so far as to suggest that Wikipedia adopt the same system. I'd enable ads for them, too.
But the guy is a total hypocrite! He talks about giving users what they want, and as a user, I want ads on Craigslist! I like them, and I'd like to know that they're earning money from my traffic, rather than being drained by it. So where is the little, unobtrusive button at the bottom of the page that says, "support Craigslist by enabling ads?" I'm looking for it, and I don't see it.
If they wanted to do this, and give most of the money to charity, that would be even cooler.
Hmmm... so, here's an anti-business model somebody could try. Create the "Progressive Advertising Network", which would be strictly opt-in by users. People who wanted to see the ads would click an enable button, and they'd show up on every page that had them. The ads would be a mixture of ads for non-profit groups and standard ads which subsidized the non-profit ads. Excess profit would be donated to some reasonably noble cause.
You misunderstand. They're not turning down money by not charging for user ads. They're turning down money by not putting ads on their website.
In response to your conclusion: Big business + the ability to pass legislation benefitting yourself at the expense of the general public does equal corruption.
Your claim that incumbency is merely a matter of name recognition is demonstrably false. Incumbents also get the vast majority of campaign contributions (especially from business). Incumbents also have the free publicity that comes from being a member of Congress. Incumbents have a great deal more sway with the people in their state who draw up district boundaries. The first one is the most important, because the candidate who spends the most usually gets the most votes.
Nor do your arguments against bankruptcy laws make much sense. "Do you really think it's 'fair' that people be allowed to absolve their own irresponsibility simply by declaring bankruptcy?" Frankly, yes.
1) Credit card companies frequently make stupid loans to high-risk individuals, with the intention of letting them run up debts that will be very difficult to repay.
2) Most people who file for bankruptcy aren't doomed by their "personal irresponsibility," but by a sudden change in their life situation, such as divorce or unexpected medical bills. Oh, wait, divorce is de facto proof of personal irresponsibility. Go ahead and say it. I'm sure you're thinking it.
3) At the time the law was passed, the lending companies that benefitted were already posting record profits.
A corporation is not a person, and shouldn't have the same powers to petition the government that normal people do. If you can't see the problems that the alternative causes, you're trying very, very hard to avoid looking for them.
Corruption is very much a part of the American system. It's not so egalitarian as the corruption in most countries, where local officials and low-level functionaries get in on it. But in the highest levels of government, if you're a corporation, you can come to a senator with a bundle of "contributions from our supportive workers", and then come to him again with a piece of legislation that conveys the benefit of a government granted monopoly, or a government contract.
Not only is this sort of thing common and destructive to our political process, it's often the best investment an industry can make. Before September 11th, the airline industry made millions in political contributions to influential senators. After September 11th, they got their payout: Billions of dollars to "save the airlines." Some of the money went to beefing up airport security (which, in retrospect, would have been a good investment, and hence the free market should have automagically taken care of it). But most of it went to bail out investors. Little of it went to saving jobs, and the airlines cut tens of thousands of jobs in the name of "efficiency".
More recently, we have a bankruptcy reform bill pretty much designed from the ground up by the credit card industry, an energy bill designed to funnel billions of dollars to the energy industry in exchange for basically doing what they've always been doing, and a "reform" of the Endangered Species Act designed solely to gut the intent of the act and promote the interests of land developers.
And guess what? It is inherent in the system, not the people. If we stepped in today and forced public financing of elections, corporations couldn't use campaign contributions as leverage to get their way. Period. It would also lower the current un-democratic incumbency rates, making long-term investments in individual politicians much more risky. Or would you like to try and argue that our 98% re-election rate just shows that our electorate always chooses the best person for the job the first time?
Yes, there is such a thing as human nature. It has good in it, and it has evil in it. Systems can be designed to promote the best in human nature, or they can be designed to subvert good intentions in favor of raw, consuming greed. We're currently living in the latter sort of system, and until we fix that system, we can try electing better people, and they'll get chewed up and spit out by it.
One word: Monopoly.
I'd explain in detail, but I'm already over my troll-feeding limit.
In retrospect, my sarcasm detector probably went home before I did. Apologies.
But what if (just toy with the idea for a moment or two) Vista isn't the most secure OS ever? Call me crazy, but I think there may be a tiny possibility that Microsoft is saying positive things about its products that may not be true.
C'mon, what are they going to say? "Vista is basically less secure than XP was, but we try to compensate for that by popping up lots and lots of messages." People wouldn't exactly be lining up to buy that, would they?
It's asinine to address Bill directly. I mean, Bill Gates kicking back and relaxing by reading Slashdot? Sure, he's a geek, but I still don't see it. Also, in the unlikely event that he is reading, what were the odds that he would let random comments from the peanut gallery "deter him for one second?" Your accolades, like the catcalls of other posters, are easy enough to brush off.
I don't think Bill Gates will be looked back upon as a "great man". While he was certainly an important force shaping our world, he got that influence by a combination of business skill, ruthlessness, and dumb luck. His net effect on technology might have been positive, or it might have led the whole system down a blind alley filled with limited choices and badly interoperating proprietary technologies.
He'll always have his supporters and his detractors, and his historical legacy will depend on whose interests the history books happen to be serving at the moment. Since they often serve wealth, I imagine that overall he will be treated well.
Standard answers:
1) That's about 1/6th of his net worth. In some ways, it's a less impressive gesture than if somebody of far lesser means gave away 1/6th of his net worth. After all, how many people could give away 98% of their wealth and still be in the billionaire's club?
2) It's a long-running debate whether such vast collections of wealth can ever be truly "earned".
3) It's a long-running argument whether Microsoft got its money in ways that promoted the overall technology ecosystem or hindered it.
4) The social inequities these vast accumulations of wealth represent are the root of many of the ills in our society.
I don't think this will help with "rapid prototyping," since it has to add an additional compilation step to your change, test, change cycle.
Nor do I think most projects will be able to depend entirely upon it. Looking at the limitations listed in the announcement, it seems like it takes away many of the things that make Python engaging as a language. What seems to remain? Imagine taking a compiling C program, translating it (as literally as possible) into python, and then let this program translate it into machine code for you.
I'm probably overstating the case. There will be certain subsets of code that might benefit from this (where people need the performance, and are willing to give up architecture independence to get it). But even if this thing matures greatly (and I'd like to see that happen), I don't think it will become ubiquitous.
Find it here. Google is our friend.
Please step back and try to see the point I've been trying to make this whole discussion: If you don't know the algorithm that will get you the right results (AI), or your source data isn't fine-grained and accurate enough (weather simulation), then all the speedup in the world is only going to get you the wrong answers faster.
This is true even for neural networks and genetic algorithms. For any given neural net configuration and set of inputs, you can only train it so much before your results stop improving (and eventually start getting worse due to overfitting). If your NN isn't properly configured (not enough layers, or an inappropriate number of nodes for each layer) or if you've chosen the wrong features as input, or if your training data is warped, the NN won't get good results. All faster hardware can do is get you to "the best you can do with what you have" more quickly.
Of course, faster processors let you do things like use genetic algorithms to select the ideal configuration for the neural net. But even genetic algorithms aren't magic. They're just one possible shortcut for exploring a large number of hypotheses. At best, GA can only match brute force of all enumerated hypotheses when it comes to the final performance of the selected algorithm. You still have to design your hypothesis space in such a way that a good solution lies within that space.
The problems I was referring to strike me as the sort where we haven't developed ideal approaches to the problem, and running the current approaches on beefier hardware aren't going to close the gap. Hence, my indignant response. My hunch is that, once we do find the proper algorithms for them, many of them will run surprisingly well on current hardware.
Fair enough. But C does provide about as little abstraction as one can expect from a language. I would argue (with some expectation of objections) that the greater the amount of abstraction, the more functionality a programmer can readily keep in his head. Say you have a subroutine that you can do the simple, slow, easily understood way, or do it the blazing-fast, well optimized, difficult to comprehend way. Which one you go for depends on several things, one of which is, "how likely is it that this code will need to be modified in the future?" If it's nice and modular, and you can wall it off so nobody need touch it again (and the performance boost is worth it) then go ahead, and shut the door behind you.
On the other hand, if it will require future programmers to understand and modify it, and you still need the performance boost, then the additional inherent complexity the programmers will have to memorize necessarily comes at the expense of something else.
Therefore, I believe that a simple, easily-grokked codebase is important. I think interpreted languages (esp. my favorites, Python and Ruby) allow for such simplicity. It might not be an overriding factor, depending on the job. But it's something to account for.
Exactly. The advantage of better hardware is the ability to either do bigger jobs, or do them in a more effective way (using better algorithms, for example). The problems described by the original poster are, for the most part, resistant to brute force (at least on the hardware we can reasonably expect in my lifetime). In many of the cases he proposes, it's not even certain what algorithm we'd throw at it if somebody dropped a Pentium 10 300-PetaHertz machine in our laps.
In your case, the key words are "real time". You knew exactly what algorithms would do the job, and you just needed a relatively trivial increase in speed to make it happen.
No. What you describe is an entirely different piece of software. Instead of running dog_brain.exe, you're running whole_world_to_find_ideal_dog_brain_via_genetic_al gorithm.exe. I'm talking about taking the same algorithm, and simply running it on faster hardware. If your text parsing system relies on the Viterbi Algorithm to classify words, then faster hardware means only that you can tackle bigger projects with the same level of accuracy. It doesn't magically turn your program into a perfect parser.
Okay, a more obvious example. Take the TI-85, the calculator I used in high school. Speed it up by a factor of a million. It doesn't get any more accurate, because it's still using the same number of bits to calculate. It doesn't magically gain the ability to solve problems that it couldn't before (unless the problem are time or memory-bound). The advantage to hardware improvements come from the ability to use the right algorithm, instead of hacks to bring in an answer in an acceptable amount of time.
Okay, I should have thrown out a few caveats about algorithmic complexity. Pure lookahead is indeed ideal, given sufficient time to explore *all* options. But pure lookahead to a depth of n? We don't know how deep a pure lookahead algorithm would have to go to beat, say, Kasparov. Might be a new way to rate chess players.
No human is going to wait around (or even live) long enough for his computer opponent to completely map out the entire chess game space, and given the branching involved, speeding up by a factor of a thousand only gives you a few more looks ahead. It's safe to assume that hardware will never be fast enough to fully map out chess space.
My point is only that, once hardware reaches a certain speed, its speed ceases to be the primary constraint, and we're left with the best performance a given algorithm can give us. You seemed to be under the impression that the solution to any AI problem was to throw more hardware at it.
I was right with you until you started gushing about the beauty of assembly. You might be able to write a single, heavily-used subroutine in assembler, and make a thing of beauty out of it. But the moment you try to do something truly ambitious, you abso-friggin-loutely need layers of abstraction. CPU cycles are one thing you don't have an unlimited amount of. Brain cycles are another.
It wasn't all that absurd. I was just saying that *if* the tax brackets cause the sort of massive disincentives that neocons claim, *then* it's absurd to put the strongest single change in incentives that close to the average person's earning power. I fully agree, nobody tries to make less in order to stay under the cap.
You seem to be under the impression that these problems you cite display inadequacies in the hardware, rather than the software. But, in the words of some fictional professor from a book I can't remember: "If you speed up a dog's brain by a factor of a million, you'll have a machine that takes only three nanoseconds to decide to sniff your crotch." Given the current software and algorithms available, more computing power alone wouldn't solve any of the problems you describe.
I really do try to be more careful in my language than that. I meant to say that there is little motivation for the departing employee to train his replacement well.
Sorry, snarky Marxist Bryce woke up today. But it does strike me as rather paternalistic, the idea that we have a few bold titans running around, and that the value of their hard work and bold decisions somehow justify a salary five hundred times that of their average employee.
I've seen individuals make good decisions, and I've seen individuals make bad decisions. I've seen committee work where a great idea was dumbed down to worthlessness, and I've seen the same process defang or eliminate really bad ideas, and substantially improve good ones. On average, I think committees perform very well, so long as their constituents are motivated by an actual interest in the subject matter and decision-making process, rather than a desire to sound creative, or just to advance their careers by being on a committee. How people were chosen for what committee work--and how they are rewarded for that work--is something that the employees would come up with at the beginning, and would be subject to tweaking.
Perhaps in some cases, people would prefer a committee of one. One person would be empowered to make certain decisions. In that case, the only difference between their operation and that of a normal company would be the fact that under worker rule, the decision maker could be recalled.
Your "bad committee decision" anecdote is certainly important. But it sounds like a situation where everyone had incentive to learn just enough to cast an "informed vote", with nobody having the extra incentive to dig deeply into the issue and present the findings to the voters. So while the bad decision may stem from the details of your process, and may not apply to all committee-style decision making.
Your assertion that we are designed to break ourselves up into leaders and followers is incorrect, or at least requires some huge caveats. You can look at baboon societies and say, "Yeah, you can always find a strict heirarchy," and draw your conclusions from that. Or you can look at the studies showing that baboons lower down the pecking order have significantly higher stress, worse health, and shorter lifespans. I honestly believe that, when we decide that a few leaders get all the power, and the rest of us simply follow their decisions, that we subject the vast majority of human beings to the same effects: higher stress, worse health, and shorter lifespans. Most any study on the subject of wealth and health will tell you that poverty-level wages are little more than a slow death sentence. But they also show that many far poorer--but far more egalitarian--countries than ours have better health outcomes than we do.
Last point, then I have to run: You ask, "If everyone is busy boning up on decisions, then who will be doing the work?" I think you have it 360 degrees backwards. It's like criticizing a company for spending too much energy on employee training. If employees have a deeper understanding of how the overall structure fits together, their work will be more effective. The incentive in worker-owned companies would be for knowledge to be spread broadly, because everyone has a stake in making the company better. Of course, there are also incentives for hoarding, which is why a culture of transparency would be critical for success.
I would argue that the companies doing the outsourcing aren't bearing the full economic costs of the transition, and that there are other costs that need to be factored in.
First, there is the cost to the employees of finding a new job, which is never fully reflected in severance pay. Severance pay probably only covers the amount of time the employee will be idle. Other costs include retraining the employee to fit the new job (paid by his new employer), the psychological effects of having his job terminated (paid by him, his family, and society at large), the costs of moving if the new job requires it, and the loss of community if he does have to move. This last item is important: while a loss of community has all sorts of ill effects, the expectation of losing it several more times over the course of your work history is even worse. Ask anyone raised by parents who got moved around a lot by the military.
People who expect to move away end up less trusting, less willing to invest in new relationships, less willing to participate in community functions or worry about local issues. There was a time when the rich still had some sort of shared community life with their workers, and it was a time when employers had loyalty to their employees. But now those who own the wealth--and decide how it will be used--are more distanced from the rest of us, and have no stake in that shared community life. So they'll happily move jobs to Outer Ohfuckitstan to earn an extra 1% return on their investment, without regard for the problems these moves cause the wider society.
The country that benefits from the new jobs also receives increased inequality between the well-off and the poor. India's rise has been accompanied by increased tension between wealthy urban areas and poor rural areas. Back when everyone was poor, the rural areas weren't likely to complain about not having electricity, because few did have it, and it's hard to want things you don't see anyone else enjoying. While an increase in the standard of living is wonderful, it is poisonous to a society when extreme inequalities in wealth exist. Of course, by this metric, our own society is very, very unhealthy.
There is another cost that employers do pay, but rarely fully account for when planning an offshoring: the loss of expertise. No matter how well the departing employees train their replacements (and what's their motivation to do a good job?) it will be years before the replacements can match the original expertise. Some of the knowledge lost in transition may never be rediscovered, leading to permanent inefficiencies.
No, I'm not a socialist. But I'm coming to the belief that collective ownership of wealth is a good thing. If Bank of America was owned by its employees, then their salaries would show up on the "income" side of the accounting ledger rather than the "expenses" side, and much of the motivation to outsource would evaporate. Otherwise, there is an inherent conflict of interest between management and labor: management wants to keep costs low, and labor is simply another expense to be minimized.
Additionally, a worker-owned setup is necessarily more efficient. Workers have the greatest incentive to make the company run as efficiently as possible, so management doesn't have to spend its time figuring out how to best motivate the employees. The workers would just tell them. Worker-owned companies are also more concerned with long-term stability, because there is no golden parachute waiting if they can just keep the stock prices from tanking in the next three quarters. There is only the promise of continued employment. Finally, when a few investors own a company, the goal is to maximize their profit. When workers own a company, they still want to maximize their returns, but they can accept a wider variety of payment: a cleaner environment, more free time to spend with their families, improvements in the social fabric of their communities, less stressful working conditions.
In short, collective ownership allows the wealth of a company to serve the many, not the few. So bring it on, free market boy.
One year's gross income? For a house? Fixer-upper or no, that's quite a find. Was it a foreclosure? Real estate market depressed in your area? Send Guido and Nunzio over to negotiate the sale?
Seriously, though. I admire the way you're living, and your ability to opt out of the overwork-to-overconsume treadmill. Most people end up getting into a house bigger than they can afford, and spend their whole lives busting ass to afford the house, so that they can say their houses are as big as the neighbor's houses. The neighbors, in turn, are spending their whole lives busting ass to pay off houses they can't afford either. But the pressure to "keep up with the Joneses" disappears once you stop seeing it as a competition.
Of course, this rat-race mentality leads to bigger and bigger houses, which makes the market unaffordable for everyone, even those who only want a place to live, rather than a status symbol. This glut of overbuilt housing isn't going away, and there are psychological and economic reasons for thinking that this housing bubble may never really "pop". So one option I never really see mentioned: splitting a house with another family. Most suburban houses are filled with crap of marginal utility, as well as copies of stuff their neighbors have. Combine households, and much of the stuff you own becomes redundant.
Other advantages: Shopping for groceries and cooking for eight doesn't take nearly twice the time of doing it for four. Only one lawn to mow. Kids have greater access to playmates and homework help. If both of you are two car families, there are probably ways to do without the fourth car. One Internet connection with twice the bandwidth is much cheaper than two separate connections. You can drop one of your cable subscriptions (but heck, why not drop both?).
Of course, there are downsides. The number of relationships to be managed goes up drastically. People may feel crowded in (though a lot of these new houses are positively huge). Splitting up the household could be messy if things go sour. But real estate has long since departed the realm of affordability for most people, especially in areas that don't require an hour or so of commuting to work. This is driving all sorts of non-American-Dream behavior, like the sudden rise in adults living with their parents. So I think that, as time goes on, more people will be considering shared living arrangements.