Private prisons. They lobby for anything that results in higher and longer incarceration rates.
Public prisons, too. The state can also extract value from its prisoners through indentured servitude. It's a (huge) net cost for the state, but there are plenty of beneficiaries in the public and private sector (because that prisoner labor is purchased by private companies). So there's plenty of financial interest in maintaining the huge US prison population even outside the morally-bankrupt private-prison industry.
Also, of course, there's always the risk that debunked bad forensics (like the FBI's ammunition divination) and improved valid forensics will exonerate prisoners, and then it's lawsuit city for the state.
Angela Davis warned us about the "Prison-Industrial Complex" in, what, 1974? Four decades on it's only grown worse. Just look at how sentence terms have increased since the mid-70s, for example.
But then it's hardly news that Jeff Sessions is a vile human being.
The office may be where you network. I work from home, aside from rare (around once a year) trips to one of the offices; but I have 2-3 hours of conference calls with coworkers every day, and the same in email, and often one-on-one calls or IM conversations. I'm highly visible, even if I'm not physically present at any of the offices.
It's easy to network remotely when you make yourself indispensable by solving a lot of problems and knowing a lot of useful stuff, and possessing reasonable social and communication skills so it's not burdensome for your coworkers to reach out.
Oh, and I'm on the cusp of my third decade as a professional developer, and only adding to my responsibilities and areas of expertise. And there are plenty of folks here with 10+ years on me. But then I work for a firm that's known for retaining expertise.
I’m just guessing that what’s now called “deep learning” is really just old-school neural networks with a spiffy new coat of marketing paint. Yes?
No. Well, depending on how precise you want to be, then perhaps "yes" in a very loose sense.
First, note that "old-school neural networks" is a very vague description. What sort of neural networks? Perceptrons? SAMs? Recurrent ANNs? Convnets? All of the above? That last is a rather large and diverse category; it's a bit like saying "hey, these new hybrid cars are just a fancy label slapped on the good old category 'automobiles'".
DL is probably best described as a subfield within the field of neural networks. DL networks are usually stacks of convolutional ANNs (convnets), plus auxiliary layers for data normalization, pooling, softmax post-processing to get a useful result, and often hidden non-convnet ANN layers such as fully-connected perceptron layers that help reduce localization. The most pertinent feature of the original DL architectures was a stack of convnets which recognize progressively more complex features, but there's a lot of other stuff going on in there too.
And, critically, before 2000 or so most researchers didn't have the computing resources available to run these sorts of deep networks. They could only speculate on what they might do, or at best simulate them by iterating over different networks, on very small (by today's standards) data sets.
And since DL became hip, there's been a ton of research, which has introduced all sorts of other devices. Before we had deep net structures we couldn't have things like bypass channels, for example.
So, no, really not the neural nets of, say, the '80s and '90s. Those designs (particularly for recurrent NNs) remain viable for many applications, and are still used. But DL really is quite different.
On the other hand, you're broadly correct in that neural nets in general - like other algorithm families with hidden parameters, such as Hidden Markov Models - are characterized by the fact that they have hidden state. It's nothing new, and the article is making a big deal out of something that's not even slightly surprising to anyone who understands the topic.
It may be relatively complex, but neural networks aren't all THAT complex. Usually there are a few hundred nodes
Certainly true for some ANN architectures, but by no means all. Simonyan & Zisserman's Very Deep network architecture has 25 layers and tens of thousands of nodes; just the three hidden fully-connected layers have over nine thousand nodes. The Inception architecture also generally employs large deep convnets - GoogLeNet has 27 layers and an entertainingly complicated structure - even though it's designed for efficiency.
These networks are typically applied to problems such as image feature extraction and image generation.
It's been done, and then some. There was that demonstration a little while back of networks developing a code, for example; that's two ANNs learning to model one another. Or take a look at Generational Adversarial Networks, for a more complex example of multiple-network systems.
An intelligent program so complex that it's almost imposible to explain or understand is in my view the correct path
I don't know what "intelligent" is supposed to mean when applied to a computer program, but: the vast majority of non-trivial software systems are "almost impossible to explain or understand". Combinatorial explosion in the state space as programs interact very quickly makes them intractable to analysis. Software pretty much is defined by what it does.
There's a very small class of software that's constructed with provable formal systems, but it's a tiny majority, and it's debatable how well even those systems could be said to be understood.
But as I noted in another post, this is far and away the natural state of things anyway. Nearly everything we deal with is intractable to analysis, except as highly simplified approximations. The great triumph of engineering has been to take tractable areas of mathematics and apply them to simplifications of intractable real-world systems with sufficiently high probability of approximately correct results that they're usable.
(Even mathematics is mostly intractable, as Chaitin demonstrated decades ago with AIT: most mathematical truths are incompressible. We usually only care about the interesting ones, which are a subset of the compressible ones, so we don't normally notice. But tractable mathematics is a vanishingly small subset of all mathematics.)
I just don't have any faith in a system that is not fully understood.
If you confine yourself to systems that are fully understood, you're not alive, nor present in the physical universe. So... good luck with that, I suppose.
On the other hand, withholding faith is in general arguably a good idea.
So I'd suggest a compromise: Don't have faith in them, but use them anyway, under an appropriate threat model that applies Perfect Bayesian Reasoning[1] when evaluating the various probabilities of failure.
[1] Or as close as you can get. Which isn't very close, for a human, but at the moment you don't have any other implementation choices.
There's the problem: if you have a trained AI and not some sort of expert system based on a collection of human knowledge it's nearly impossible to say how it will handle the unexpected near-garbage input.
It's not a problem. Whoever wrote that piece for MTR shouldn't be in the science-writing business.
We deal with analysis-intractable systems all the time. They're the vast majority of the systems we deal with, in engineering and in everyday life. Most of the physical systems of the car aren't tractable to analysis. Weather isn't. Human health isn't. The electrical grid isn't. Animal behavior isn't.
We have plenty of computer-controlled industrial systems that have intractable control mechanisms. Some are famous - the Sendai train system with its fuzzy-logic control system, for example. Large fuzzy-logic systems are very similar to deep convolutional ANNs in this regard - they coalesce weighted inputs with a nonlinear rectifying function. (I think the Sendai train system uses MAX rather than functions like ReLU or tanh, which are common in ANNs, but it's the same principle.)
So we do what we do with other intractable systems: simulation, empirical study, and statistical analysis to derive a statistical-mechanical model. The Sendai train system underwent years of simulated testing, with hundreds of thousands of simulations (according to an interview with Zadeh in DDJ some years back). Claiming that we need an exact, compact model of the system in order to declare it fit for purpose is naive, in terms of both engineering and history.
Why is it that people who don't watch TV are so damned proud of it that they need to announce it every time TV is mentioned?
It's the veganism of mass-media entertainment.
I'm a curmudgeon myself, but, really, folks - find something more interesting to haughtily reject. Proclaiming your independence from broadcast television is like proudly announcing your stand against eating kale.
Now, if you want to tell us all how you never use emacs...
Oh, get off your high horse. I have no great fondness for television. I rarely watch it, except in social situations - and only then if the other people in the room are actually paying attention to it. But no moral points are awarded for dismissing it.
I've seen television programs that are excellent works of art, just as I've read books that were inane wastes of paper. I've gone to parks and flown kites and taken pictures and played with children. I've knapped flint and painted houses and carved antler and plumbed bathrooms and written poetry. All of these can be worthwhile activities, and with some effort any of them can be rendered trite, dull, and meaningless, too.
Some people care about television programming and the writers' strike, just as some people care about professional sports, or the declining number of speakers of Gaelic, or space exploration. That is not a personal failing on their part, and exhorting them not to care only demonstrates a narrowness in your conception of human existence. And the same goes for your cheerleaders below.
Well, that's hardly fair. Try walking a few miles through a recently-harvested corn (maize) field, for example. Or over hot lava.
Dude did say he was in South Florida, so he may have been attacked by an alligator while walking. Or run over by a retiree. Or maybe there was a hurricane.
Of course, when I were a lad, we had to put up with such things on our daily trek to school. And did we complain? Well, yes, but mostly about having to work in the salt mine after class.
For that matter, there are assembly languages which don't target a real physical CPU.
There's Knuth's MIX, for example; the (original) target for that is a CPU which is either imaginary or in the mind of the reader, depending on what ontological hairs you care to split.
Or there's the MI ("Machine Instruction") language for the AS/400, which looks like assembly and works like assembly, but targets a notional CPU which is not, in fact, any of the actual CPUs that the '400 and its successors were implemented with. But it was still an assembly language, in that it directly specified the operations of the notional machine.
You just need to interpret Pai's comments in the proper context. In his case, the context is "I have always been, and will always be, the lackey of the telecom industry". Just look at his resumé.
So when he claims there's adequate competition in the "broadband" ISP market, what he means is "ha ha, fuck you, customers!".
1. The big screen. There's something to be said about watching visual storytelling on a three-story screen, particularly when the film really takes advantage of the format.
If there's something to be said for it, then say it.
2. People everywhere. A group of people laughing together simultaneously triggers a feeling that you should laugh, too; during a suspenseful moment, you can feel dozens of strangers suck in their breath together.
A group of people talking to each other and making phone calls during the film, on the other hand...
3. Focus. Even people who try their hardest to give a movie their undivided attention on a living-room screen have fallen victim to temptations like "Well, I'm just sitting here, I might as well pay the electric bill."
Ah, argument by assertion. If I want to watch something, I watch it. If I don't want to watch it, I don't.
4. Relentlessness. Part of the advantage of that kind of focus is that movies that are tense, scary, or deeply emotional can cast much more of a spell over you when you don't have the option to pause or turn away from the worst, then rewind later to catch it safely out of context.
More argument by assertion.
5. A massive speaker system.
... is an excellent reason not to go to the cinema. I might be willing to pay cash to see a film; I'm not willing to pay with my hearing.
6. Previews.
Oh, hey! We have ads! You can't get those at home!
7. Disruption. A problem with watching movies at home is that it makes the film-watching experience blur into the same experience as surfing cable channels, running a Netflix comedy show in the background while you do dishes, or half-assedly watching an Adventure Time marathon while stoned.
Assertion, assertion, assertion. Perhaps I do not do any of those things. (As it happens, I don't.)
8. Alone time. Going to the movies with friends or your significant other can be a cherished pastime, especially when you're surrounded by an excited audience.
Whereas I can't cherish the time I spend with friends and family at home? And - mirabile dictu - some of us don't think an audience improves things.
9. 32 ounces of cola in the dark.
Ugh.
10. Bragging rights.
What the ever-lovin' fuck? People brag about going to the movies?
What a mind-bogglingly stupid series of arguments.
I don't think geographical limits on reproduction count as eugenics, unless geography is just being used as some sort of proxy for genotype. Which, of course, it often is; but not I think in GP's modest proposal.
Of course that proposal is what rich cultures actually do - the native birthrate goes down as the standard of living climbs, and that higher standard of living attracts immigrants. When that process doesn't reach equilibrium, you end up with an "aging society" and dropping productivity.
I don't know where GGP troll's "housing problem" is, but it ain't the US, aside from some places that have artificially limited the housing stock (hello, Bay Area!). Even NYC isn't terribly densely populated by international standards. And the rest of the US is composed largely of great honking areas of empty space. I drive through Kansas several times a year, and there are places on I-70 where you can travel 30 miles or more without seeing a permanent structure. And that's the populous corridor in Kansas.
Or Wyoming. Wyoming's a little more than half a million people in nearly 100000 sq. miles - it's a bit bigger than the UK, with less than 1% of the population.
Of course, you can't just drop a million immigrants into the middle of nowhere; you need jobs of some sort for them, and services, and blah blah blah. But room? Room we got.
In theory, as other posters have noted, they could make the market more efficient - do a better job of matching prospective tenants to landlords, and thus let landlords provide a wider variety of lease options, etc.
The value they would add in that case is increased satisfaction per unit price for the tenant (who might pay more but get a property and/or terms he or she values more highly), and/or reduced opportunity costs for tenants and landlords - because the market is more transparent, people can theoretically arrive at Nash equilibria with higher total satisfaction.
In practice? I wouldn't count on it. You never know - many people love AirBnb, for example, and basically it expanded the short-term rental market and made it more efficient. (I wouldn't touch it myself, but that's because I'm a risk-averse, opportunity-cost averse, contrarian curmudgeon. I don't want to be on your lawn.) An auction for long-term housing rental seems like it could have all sorts of adverse consequences, but making economic predictions rarely works out well.
Why? I'm happy. Happy pretty much every day. Some days just hovering in the neighborhood of "content", some suffused with a cheerful glow; but happy to some extent, anyway.
Here are some things that are unlikely to make me happier: More days of vacation; I never use mine up anyway. More sick days; I hardly ever use any of those. "Worker's protection", the very idea of which makes my skin crawl - I'll do what I want, when I want, thanks anyway.
"Perfect healthcare" sounds ominous. Perhaps you don't know what the word "perfect" means?
I have a retirement plan. Yes, unemployment coverage in the US is wildly insufficient, but I wouldn't claim the US is ideal. Hell, I won't claim it's better than anywhere else, for me and certainly not for other people. But it's just fine for me.
I don't understand this weird dick-waving "try to beat that" crap. Are you happy? Great! That doesn't mean other people can't be unless they have what you have.
Slashdot Mirror
Private prisons. They lobby for anything that results in higher and longer incarceration rates.
Public prisons, too. The state can also extract value from its prisoners through indentured servitude. It's a (huge) net cost for the state, but there are plenty of beneficiaries in the public and private sector (because that prisoner labor is purchased by private companies). So there's plenty of financial interest in maintaining the huge US prison population even outside the morally-bankrupt private-prison industry.
Also, of course, there's always the risk that debunked bad forensics (like the FBI's ammunition divination) and improved valid forensics will exonerate prisoners, and then it's lawsuit city for the state.
Angela Davis warned us about the "Prison-Industrial Complex" in, what, 1974? Four decades on it's only grown worse. Just look at how sentence terms have increased since the mid-70s, for example.
But then it's hardly news that Jeff Sessions is a vile human being.
The office may be where you network. I work from home, aside from rare (around once a year) trips to one of the offices; but I have 2-3 hours of conference calls with coworkers every day, and the same in email, and often one-on-one calls or IM conversations. I'm highly visible, even if I'm not physically present at any of the offices.
It's easy to network remotely when you make yourself indispensable by solving a lot of problems and knowing a lot of useful stuff, and possessing reasonable social and communication skills so it's not burdensome for your coworkers to reach out.
Oh, and I'm on the cusp of my third decade as a professional developer, and only adding to my responsibilities and areas of expertise. And there are plenty of folks here with 10+ years on me. But then I work for a firm that's known for retaining expertise.
I’m just guessing that what’s now called “deep learning” is really just old-school neural networks with a spiffy new coat of marketing paint. Yes?
No. Well, depending on how precise you want to be, then perhaps "yes" in a very loose sense.
First, note that "old-school neural networks" is a very vague description. What sort of neural networks? Perceptrons? SAMs? Recurrent ANNs? Convnets? All of the above? That last is a rather large and diverse category; it's a bit like saying "hey, these new hybrid cars are just a fancy label slapped on the good old category 'automobiles'".
DL is probably best described as a subfield within the field of neural networks. DL networks are usually stacks of convolutional ANNs (convnets), plus auxiliary layers for data normalization, pooling, softmax post-processing to get a useful result, and often hidden non-convnet ANN layers such as fully-connected perceptron layers that help reduce localization. The most pertinent feature of the original DL architectures was a stack of convnets which recognize progressively more complex features, but there's a lot of other stuff going on in there too.
And, critically, before 2000 or so most researchers didn't have the computing resources available to run these sorts of deep networks. They could only speculate on what they might do, or at best simulate them by iterating over different networks, on very small (by today's standards) data sets.
And since DL became hip, there's been a ton of research, which has introduced all sorts of other devices. Before we had deep net structures we couldn't have things like bypass channels, for example.
So, no, really not the neural nets of, say, the '80s and '90s. Those designs (particularly for recurrent NNs) remain viable for many applications, and are still used. But DL really is quite different.
On the other hand, you're broadly correct in that neural nets in general - like other algorithm families with hidden parameters, such as Hidden Markov Models - are characterized by the fact that they have hidden state. It's nothing new, and the article is making a big deal out of something that's not even slightly surprising to anyone who understands the topic.
It may be relatively complex, but neural networks aren't all THAT complex. Usually there are a few hundred nodes
Certainly true for some ANN architectures, but by no means all. Simonyan & Zisserman's Very Deep network architecture has 25 layers and tens of thousands of nodes; just the three hidden fully-connected layers have over nine thousand nodes. The Inception architecture also generally employs large deep convnets - GoogLeNet has 27 layers and an entertainingly complicated structure - even though it's designed for efficiency.
These networks are typically applied to problems such as image feature extraction and image generation.
It's been done, and then some. There was that demonstration a little while back of networks developing a code, for example; that's two ANNs learning to model one another. Or take a look at Generational Adversarial Networks, for a more complex example of multiple-network systems.
The biggest problem with understanding neural networks is convincing people there's no need to understand neural networks.
Deep convolutional ANNs are examples of Bayesian statistics to pretty much the same extent that they're examples of Euclidean plane geometry.
Now, if we were talking about Hidden Markov Models or MEMMs, say, you might have a point. But DL ANNs? That is really a stretch.
But thanks for playing.
An intelligent program so complex that it's almost imposible to explain or understand is in my view the correct path
I don't know what "intelligent" is supposed to mean when applied to a computer program, but: the vast majority of non-trivial software systems are "almost impossible to explain or understand". Combinatorial explosion in the state space as programs interact very quickly makes them intractable to analysis. Software pretty much is defined by what it does.
There's a very small class of software that's constructed with provable formal systems, but it's a tiny majority, and it's debatable how well even those systems could be said to be understood.
But as I noted in another post, this is far and away the natural state of things anyway. Nearly everything we deal with is intractable to analysis, except as highly simplified approximations. The great triumph of engineering has been to take tractable areas of mathematics and apply them to simplifications of intractable real-world systems with sufficiently high probability of approximately correct results that they're usable.
(Even mathematics is mostly intractable, as Chaitin demonstrated decades ago with AIT: most mathematical truths are incompressible. We usually only care about the interesting ones, which are a subset of the compressible ones, so we don't normally notice. But tractable mathematics is a vanishingly small subset of all mathematics.)
I just don't have any faith in a system that is not fully understood.
If you confine yourself to systems that are fully understood, you're not alive, nor present in the physical universe. So ... good luck with that, I suppose.
On the other hand, withholding faith is in general arguably a good idea.
So I'd suggest a compromise: Don't have faith in them, but use them anyway, under an appropriate threat model that applies Perfect Bayesian Reasoning[1] when evaluating the various probabilities of failure.
[1] Or as close as you can get. Which isn't very close, for a human, but at the moment you don't have any other implementation choices.
There's the problem: if you have a trained AI and not some sort of expert system based on a collection of human knowledge it's nearly impossible to say how it will handle the unexpected near-garbage input.
It's not a problem. Whoever wrote that piece for MTR shouldn't be in the science-writing business.
We deal with analysis-intractable systems all the time. They're the vast majority of the systems we deal with, in engineering and in everyday life. Most of the physical systems of the car aren't tractable to analysis. Weather isn't. Human health isn't. The electrical grid isn't. Animal behavior isn't.
We have plenty of computer-controlled industrial systems that have intractable control mechanisms. Some are famous - the Sendai train system with its fuzzy-logic control system, for example. Large fuzzy-logic systems are very similar to deep convolutional ANNs in this regard - they coalesce weighted inputs with a nonlinear rectifying function. (I think the Sendai train system uses MAX rather than functions like ReLU or tanh, which are common in ANNs, but it's the same principle.)
So we do what we do with other intractable systems: simulation, empirical study, and statistical analysis to derive a statistical-mechanical model. The Sendai train system underwent years of simulated testing, with hundreds of thousands of simulations (according to an interview with Zadeh in DDJ some years back). Claiming that we need an exact, compact model of the system in order to declare it fit for purpose is naive, in terms of both engineering and history.
So many brave refusniks telling everyone else how much better their television-less lives are! Truly, Slashdot is the locus of enlightenment.
Then I watched the difference...
This is nearly a perfect argument. "Based on a hazy claim of anecodotal observation, I posited a correlation and therefore deduced causation."
Really, the only way you could improve that is by imputing it to supernatural forces.
Why is it that people who don't watch TV are so damned proud of it that they need to announce it every time TV is mentioned?
It's the veganism of mass-media entertainment.
I'm a curmudgeon myself, but, really, folks - find something more interesting to haughtily reject. Proclaiming your independence from broadcast television is like proudly announcing your stand against eating kale.
Now, if you want to tell us all how you never use emacs...
I'm going to give you one word that will blow your mind...
... it is the thing that I like, and everyone else will like it too, because everyone is the same as me.
Oh, get off your high horse. I have no great fondness for television. I rarely watch it, except in social situations - and only then if the other people in the room are actually paying attention to it. But no moral points are awarded for dismissing it.
I've seen television programs that are excellent works of art, just as I've read books that were inane wastes of paper. I've gone to parks and flown kites and taken pictures and played with children. I've knapped flint and painted houses and carved antler and plumbed bathrooms and written poetry. All of these can be worthwhile activities, and with some effort any of them can be rendered trite, dull, and meaningless, too.
Some people care about television programming and the writers' strike, just as some people care about professional sports, or the declining number of speakers of Gaelic, or space exploration. That is not a personal failing on their part, and exhorting them not to care only demonstrates a narrowness in your conception of human existence. And the same goes for your cheerleaders below.
I think all state supreme justices are not elected
The Michigan Supreme Court, for one, is made up of elected judges, except for interim appointments.
Well, that's hardly fair. Try walking a few miles through a recently-harvested corn (maize) field, for example. Or over hot lava.
Dude did say he was in South Florida, so he may have been attacked by an alligator while walking. Or run over by a retiree. Or maybe there was a hurricane.
Of course, when I were a lad, we had to put up with such things on our daily trek to school. And did we complain? Well, yes, but mostly about having to work in the salt mine after class.
I've encountered plenty of cabbies that were not creepy at all, so unless all Uber drivers have negative creepiness, your claim is a load of bullshit.
For that matter, there are assembly languages which don't target a real physical CPU.
There's Knuth's MIX, for example; the (original) target for that is a CPU which is either imaginary or in the mind of the reader, depending on what ontological hairs you care to split.
Or there's the MI ("Machine Instruction") language for the AS/400, which looks like assembly and works like assembly, but targets a notional CPU which is not, in fact, any of the actual CPUs that the '400 and its successors were implemented with. But it was still an assembly language, in that it directly specified the operations of the notional machine.
You just need to interpret Pai's comments in the proper context. In his case, the context is "I have always been, and will always be, the lackey of the telecom industry". Just look at his resumé.
So when he claims there's adequate competition in the "broadband" ISP market, what he means is "ha ha, fuck you, customers!".
1. The big screen. There's something to be said about watching visual storytelling on a three-story screen, particularly when the film really takes advantage of the format.
If there's something to be said for it, then say it.
2. People everywhere. A group of people laughing together simultaneously triggers a feeling that you should laugh, too; during a suspenseful moment, you can feel dozens of strangers suck in their breath together.
A group of people talking to each other and making phone calls during the film, on the other hand...
3. Focus. Even people who try their hardest to give a movie their undivided attention on a living-room screen have fallen victim to temptations like "Well, I'm just sitting here, I might as well pay the electric bill."
Ah, argument by assertion. If I want to watch something, I watch it. If I don't want to watch it, I don't.
4. Relentlessness. Part of the advantage of that kind of focus is that movies that are tense, scary, or deeply emotional can cast much more of a spell over you when you don't have the option to pause or turn away from the worst, then rewind later to catch it safely out of context.
More argument by assertion.
5. A massive speaker system.
... is an excellent reason not to go to the cinema. I might be willing to pay cash to see a film; I'm not willing to pay with my hearing.
6. Previews.
Oh, hey! We have ads! You can't get those at home!
7. Disruption. A problem with watching movies at home is that it makes the film-watching experience blur into the same experience as surfing cable channels, running a Netflix comedy show in the background while you do dishes, or half-assedly watching an Adventure Time marathon while stoned.
Assertion, assertion, assertion. Perhaps I do not do any of those things. (As it happens, I don't.)
8. Alone time. Going to the movies with friends or your significant other can be a cherished pastime, especially when you're surrounded by an excited audience.
Whereas I can't cherish the time I spend with friends and family at home? And - mirabile dictu - some of us don't think an audience improves things.
9. 32 ounces of cola in the dark.
Ugh.
10. Bragging rights.
What the ever-lovin' fuck? People brag about going to the movies?
What a mind-bogglingly stupid series of arguments.
I thought Ace Ventura had a different job. Something about pets?
Perhaps one of them was a Commodore PET.
I don't think geographical limits on reproduction count as eugenics, unless geography is just being used as some sort of proxy for genotype. Which, of course, it often is; but not I think in GP's modest proposal.
Of course that proposal is what rich cultures actually do - the native birthrate goes down as the standard of living climbs, and that higher standard of living attracts immigrants. When that process doesn't reach equilibrium, you end up with an "aging society" and dropping productivity.
I don't know where GGP troll's "housing problem" is, but it ain't the US, aside from some places that have artificially limited the housing stock (hello, Bay Area!). Even NYC isn't terribly densely populated by international standards. And the rest of the US is composed largely of great honking areas of empty space. I drive through Kansas several times a year, and there are places on I-70 where you can travel 30 miles or more without seeing a permanent structure. And that's the populous corridor in Kansas.
Or Wyoming. Wyoming's a little more than half a million people in nearly 100000 sq. miles - it's a bit bigger than the UK, with less than 1% of the population.
Of course, you can't just drop a million immigrants into the middle of nowhere; you need jobs of some sort for them, and services, and blah blah blah. But room? Room we got.
What benefit do you think this company provides?
In theory, as other posters have noted, they could make the market more efficient - do a better job of matching prospective tenants to landlords, and thus let landlords provide a wider variety of lease options, etc.
The value they would add in that case is increased satisfaction per unit price for the tenant (who might pay more but get a property and/or terms he or she values more highly), and/or reduced opportunity costs for tenants and landlords - because the market is more transparent, people can theoretically arrive at Nash equilibria with higher total satisfaction.
In practice? I wouldn't count on it. You never know - many people love AirBnb, for example, and basically it expanded the short-term rental market and made it more efficient. (I wouldn't touch it myself, but that's because I'm a risk-averse, opportunity-cost averse, contrarian curmudgeon. I don't want to be on your lawn.) An auction for long-term housing rental seems like it could have all sorts of adverse consequences, but making economic predictions rarely works out well.
Try to beat that, US.
Why? I'm happy. Happy pretty much every day. Some days just hovering in the neighborhood of "content", some suffused with a cheerful glow; but happy to some extent, anyway.
Here are some things that are unlikely to make me happier: More days of vacation; I never use mine up anyway. More sick days; I hardly ever use any of those. "Worker's protection", the very idea of which makes my skin crawl - I'll do what I want, when I want, thanks anyway.
"Perfect healthcare" sounds ominous. Perhaps you don't know what the word "perfect" means?
I have a retirement plan. Yes, unemployment coverage in the US is wildly insufficient, but I wouldn't claim the US is ideal. Hell, I won't claim it's better than anywhere else, for me and certainly not for other people. But it's just fine for me.
I don't understand this weird dick-waving "try to beat that" crap. Are you happy? Great! That doesn't mean other people can't be unless they have what you have.