If you actually read and understood any of his robot stories, the theme was consistent: robots are no better than we program them to be and they can never be as good/smart/ethical as we are, period.
That argument is meaningless because it always works: 1903: Heavier-than-air powered flight is more tractable than we previously thought. 1957: Putting objects in orbit is more tractable than we previously thought. 1969: Landing on the moon is more tractable than we previously thought. 2016: The game of Go is more tractable than we previously thought. 20xx: Human-level cognition is more tractable than we previously thought.
Yes, a lot of people are over-optimistic about AI at any given time, but it is moving forward as evidenced by milestones such as AlphaGo. Steam power took 150 years to go from possible to common, I don't see why AGI can't be similar and still be considered a success (serious AI efforts started in the 1950s). I think we're just too impatient now, since a few aspects of technology became widespread more quickly.
To add to the above, search for "denoising autoencoder" as the classic way to do this. You can then take the learned intermediate representation and use it in a classifier.
My worry with that is that there's a big disparity in who gets it, as new construction tends to go to the upper-middle class and rich (at least in most of the US). In my previous state (Pennsylvania) the fiber really only went to the expensive neighborhoods; from a business sense I understand -- regular neighborhoods would be less likely to care about those packages, but it still leaves a bad taste. If the gov't is running something, it ought to be inclusive.
Of course what you say would be a good part of the ongoing solution, making sure the city doesn't need to spend nearly as much after the initial push.
Thanks that's a good point. I don't know much about the networks since I've never lived somewhere where fiber was a realistic option. In my previous town DSL barely worked over the ancient copper.
Mind if you share where you live, or other places with the same model? I'd love to be able to point to real examples.
If my state ever gets it's head out of it's rear and drops its municipal network quazi-ban, I'd love to push this in my city. We were briefly on the "possible" list for Google Fiber, but instead of waiting for a unicorn I feel like we'd do well to push it far enough on our own that it becomes cheap for any ISP to move in.
Which part of the 4 sentence post is a misrepresentation? The first two sentences are verifiable facts, and with obvious math you can see that 70% are R-leaning, just like the GGP "corrected" me with. The remaining two sentences which make any kind of claim are:
(1) The pattern is that of corporate corruption of politics, which affects both parties but Republicans more. (2) Since you have solidly blue states such as CA and WA in on this, you really can't call it just a Republican problem.
Which of those two do you disagree with?
In my other post I broke this down in a spreadsheet; if you can link a table with legislature majorities I'd be happy to incorporate it. Note that most of these regulations came prior to 2016, and 2008/2012 has a more D states and the population was split was 58% and 63%. Of course states don't need to be defined by who they voted for president, but that is the definition most people use for "red/blue" rather than governor or the current state legislature.
You know when you combine our pots, 30+70=100 and 7+17=24, so you're saying the same thing. Also, it's 23 states, not 24, because you counted Washington twice as it shows up in the report table twice.
You were implying this is somehow a Republican-only problem, which as a Californian, I can confirm that it's not. Since CA has 12% of the US population all by itself, and is D controlled at nearly every level, it's pretty disingenuous to call it merely cherry-picked. Hell in my post I even say that it's mostly a Republican problem ("affects both parties but Republicans more"). However if you must view everything through the lens of "if one party is wrong the other must be right", then I can't help you.
Besides, raw counts are dumb since 30/50 states voted R, so really we ought condition or even weight by population: Voted D: 6/20 = 30%, 49% population-weighted Voted R: 17/30 = 56%, 69% population-weighted
So if you are in a D state, odds are 50/50 that you have restricted municipal broadband. In R states it jumps to almost 70%. What it most definitely is not, in either case, is near zero.
p>If Republicans would stop preventing broadband competition we'd be far better off. And before anyone wants to whine about being partisan, go take a look at the places which have outlawed municipal broadband. See the pattern?
California, Colorado, Minnesota, Nevada, Washington, Virginia all voted blue in the last presidential election, and all have some form of restriction or hurdle for municipal broadband. That's about 30% of the states with such regulations. The pattern is that of corporate corruption of politics, which affects both parties but Republicans more. Since you have solidly blue states such as CA and WA in on this, you really can't call it just a Republican problem.
My dream: local governments (or the local power company) run the "last mile" passive fiber to every home. Then any company can apply to come in and start hooking up at the switch boxes. This means new offerings like Google Fiber could hook up quickly, and the old guard can still provide competitive service if they choose to (also dragging them into a fiber-first model). No need to fret over who gets connected at the house level, because you have public oversight at that level, and not having to do the last-mile means there's less incentive to hook up only the rich neighborhoods, because all of them can be done fairly efficiently once you have backhaul. This design also keeps the government from trying to be an ISP, which they aren't really equipped for -- instead they maintain the street-level infrastructure, something they do a lot of already.
Before telecom deregulation I had a small ISP over Verizon's copper, and (for the time) it was great. The ISP of course got killed off as soon as Verizon was allowed to stop sharing the lines. A decade of stagnation followed. I'd love to see the smallest changes on the public side to make private competition viable, and a municipally owned last mile makes a lot of sense.
The mere $7K fine is perhaps the real travesty here. With a fine is that low it wouldn't even be in the company's financial interest to follow it, as the average cost of downtime DIVIDED by the probability of being caught is probably way more than $7K. The fine ought to be calculated explicitly to make sure it's cheaper to follow, that way even sociopathic management will want to do the right thing.
Ok, I have a PhD relevant to doomsday scenarios (robotics and AI, from a top university).
On the Bulletin's Science and Security Board, only 8/14 have PhDs, and most of those are related to environment or international policy. They don't have any scientists in the area of AI (overblown but nonzero threat) or biological warfare or disease (generally underestimated threat).
The fact that Trump is a dipshit doesn't mean that the rest of us have to lose our minds. Many of my friends on social media have lost all ability to think or reason, and just pass through shoddy unsubstantiated articles as fact, which is sad because that's the problem with the POTUS that they are decrying in the first place. Fight idiocy with well sourced and reasoned explanations, and calm refusal to capitulate with the worst of it. Do not return in kind.
If you look at the Science and Security Board of the Bulletin, you'll see that only two of the 14 members have an atomic science background, and only two more have meaningful nuclear policy experience. This is very different from how it was when it started in 1945. Most of the people there now are environmental and public policy folks. Only 8/14 seem to have PhDs (what many would expect when you say "scientist").
It'd be more accurate to call it the Bulletin of Environmental Policy Scientists. In that lens their determination does make sense, as Trump will be nothing but bad for global warming. However a clock-to-midnight is a poor representation of a threat like that, which takes sustained and difficult work over a long period rather than a reduction of tensions to solve.
Apparently Elon Musk has tried to float the idea of a carbon tax with Trump. While unsuccessful so far, that's probably a bigger impact than the Bulletin will have during this administration.
The only "barriers" that exist are those created by Republican politicians.
Then explain why California has such a crappy broadband law. It does seem that deep Red states have the worst laws (outright bans), but the major ISPs are "friends" of every government and are just as happy with planting minefields. As long as it stops municipalities from solving the last mile, it's a good deal for them.
I don't think many people seriously believed voting itself was hacked.
Actually 50% of Clinton voters believe that, or about a quarter of the US population. Of course Trump voters had their own preferred conspiracy theories as you can see from the same report. What it shows is that people believe the narrative of their preferred news sources, and that neither major party is immune.
Rolling Stone has a nice article with an overview of the whole situation and why we should be skeptical. It's nice to see that coming from them, because MSNBC won't say it (gotta support out side) and nobody is going to believe Fox saying it, and sadly the rest of the press is starting to fall in line with one side or the other.
TFS should have included Google's response (already in TFA):
“We’ve worked hard to comply with the OFCCP’s current audit. However, the handful of OFCCP requests that are the subject of the complaint are overbroad in scope, or reveal confidential data, and we've made this clear to the OFCCP, to no avail. These requests include thousands of employees’ private contact information which we safeguard rigorously. We hope to continue working with OFCCP to resolve this matter.”
Perhaps a better approach is Google's NaCl, where intermediate code is translated more directly to native code, while putting security guarantees in place. But here, of course, cross-platform support is an issue.
I'm hoping that eventually browsers, mobile phones, and cloud hosting will just become sandboxed LLVM targets. Then people could use whatever language they like, wherever they want.
I live near Google and also work there. The bus stops are not anywhere near campus, as they are only intended for people with long commutes where pooling makes sense. I live 7 miles from my building, which is bikable and drivable, so no bus stop or bus for me. What you see around campus is empty buses during the day because they need somewhere to hang out between the morning and evening rush. I think you might be misinterpreting this as how they always look.
They are not driving them empty up and down the peninsula all day. Sometimes during rush hour there is one direction that'll be almost empty, but in those cases there are a *lot* of people the other way.
Many-to-one transportation systems (like corporate shuttles) are much easier to run efficiently than many-to-many systems (like most city buses). This is why systems use hubs -- even though they take you well out of your way, they are cheaper for the carrier. But in the case of a shuttle, the hub *is* where you want to go, so there is no loss of efficiency.
How about where a company sinks several person-years of effort into a library or software suite, then decides to open source it? If an external developer spends a few days fixing a bug, it would be nice if the company that footed most of the development cost could use that fix in their internal deployments. Only fair, right?
Now hopefully it makes sense why companies release mostly BSD software (Clang being a good example). It is also why developers BSD things that they want companies to embrace (basic components where wide use helps like zlib, or things where you'd like a company to foot the long-term cost).
I've released both GPLv2 and MIT-licensed libraries. My current employer allows me to open source some work, but strongly encourages Apache licensing. I follow that, because they paid me to write it and it seems unfair to limit their future use.
Use quotes around words, or enable "verbatim results" in the options. The + thing was misunderstood and misused by most users, and they figured experts could RTFM.
The thesis you are basing your work is from 1977; while no doubt current when it was written, there is has been a lot of work on human signal decoding since then.
I'd strongly suggest looking at Hidden Markov Models: http://en.wikipedia.org/wiki/Hidden_Markov_model While some recent methods have gone beyond HMMs for speech recognition, that's been the baseline "good" solution for the past decade.
Since this is a binary signal problem another approach to consider would be Markov Random Fields (MRFs) which could be used as an initial de-noising pass or even as a full decoder if you set the cost functions right.
Your idea of user adaptation is pretty reasonable, but my guess is the primary thing that matters would be an overall speed scaling. IOW for good decoding you probably just need to normalize the average letter rate between users.
For public stop usage the SFMTA was aware of and already working toward a solution since late 2011: https://www.sfmta.com/projects-planning/projects/shuttle-partners-program/detail It's all laid out pretty reasonably without having to get into a ticketing war or protests. These protesters are late to this issue, yet will probably claim credit when the mutli-year regulation update goes in place next year.
Of course, this is just a side issue for the bus protesters, it is more about the evictions. There are a lot of things driving that from zoning regulations to economics, so they pick a visible if somewhat poorly representative target.
In your case, it does seems like quite a traffic growth problem, but replacing the each bus with 30+ cars doesn't seem like a good solution.
Slashdot Mirror
True for the first part, but...
If you actually read and understood any of his robot stories, the theme was consistent: robots are no better than we program them to be and they can never be as good/smart/ethical as we are, period.
Read Robots and Empire.
That argument is meaningless because it always works:
1903: Heavier-than-air powered flight is more tractable than we previously thought.
1957: Putting objects in orbit is more tractable than we previously thought.
1969: Landing on the moon is more tractable than we previously thought.
2016: The game of Go is more tractable than we previously thought.
20xx: Human-level cognition is more tractable than we previously thought.
Yes, a lot of people are over-optimistic about AI at any given time, but it is moving forward as evidenced by milestones such as AlphaGo. Steam power took 150 years to go from possible to common, I don't see why AGI can't be similar and still be considered a success (serious AI efforts started in the 1950s). I think we're just too impatient now, since a few aspects of technology became widespread more quickly.
https://en.wikipedia.org/wiki/Banqiao_Dam
171,000 killed, or 40x the deaths attributable to the Chernobyl disaster.
You can say we (mostly) don't build dams like that anymore, but we don't build reactors like Chernobyl anymore either.
To add to the above, search for "denoising autoencoder" as the classic way to do this. You can then take the learned intermediate representation and use it in a classifier.
My worry with that is that there's a big disparity in who gets it, as new construction tends to go to the upper-middle class and rich (at least in most of the US). In my previous state (Pennsylvania) the fiber really only went to the expensive neighborhoods; from a business sense I understand -- regular neighborhoods would be less likely to care about those packages, but it still leaves a bad taste. If the gov't is running something, it ought to be inclusive.
Of course what you say would be a good part of the ongoing solution, making sure the city doesn't need to spend nearly as much after the initial push.
Thanks that's a good point. I don't know much about the networks since I've never lived somewhere where fiber was a realistic option. In my previous town DSL barely worked over the ancient copper.
Mind if you share where you live, or other places with the same model? I'd love to be able to point to real examples.
If my state ever gets it's head out of it's rear and drops its municipal network quazi-ban, I'd love to push this in my city. We were briefly on the "possible" list for Google Fiber, but instead of waiting for a unicorn I feel like we'd do well to push it far enough on our own that it becomes cheap for any ISP to move in.
Which part of the 4 sentence post is a misrepresentation? The first two sentences are verifiable facts, and with obvious math you can see that 70% are R-leaning, just like the GGP "corrected" me with. The remaining two sentences which make any kind of claim are:
(1) The pattern is that of corporate corruption of politics, which affects both parties but Republicans more.
(2) Since you have solidly blue states such as CA and WA in on this, you really can't call it just a Republican problem.
Which of those two do you disagree with?
In my other post I broke this down in a spreadsheet; if you can link a table with legislature majorities I'd be happy to incorporate it. Note that most of these regulations came prior to 2016, and 2008/2012 has a more D states and the population was split was 58% and 63%. Of course states don't need to be defined by who they voted for president, but that is the definition most people use for "red/blue" rather than governor or the current state legislature.
You know when you combine our pots, 30+70=100 and 7+17=24, so you're saying the same thing. Also, it's 23 states, not 24, because you counted Washington twice as it shows up in the report table twice.
You were implying this is somehow a Republican-only problem, which as a Californian, I can confirm that it's not. Since CA has 12% of the US population all by itself, and is D controlled at nearly every level, it's pretty disingenuous to call it merely cherry-picked. Hell in my post I even say that it's mostly a Republican problem ("affects both parties but Republicans more"). However if you must view everything through the lens of "if one party is wrong the other must be right", then I can't help you.
Besides, raw counts are dumb since 30/50 states voted R, so really we ought condition or even weight by population:
Voted D: 6/20 = 30%, 49% population-weighted
Voted R: 17/30 = 56%, 69% population-weighted
So if you are in a D state, odds are 50/50 that you have restricted municipal broadband. In R states it jumps to almost 70%. What it most definitely is not, in either case, is near zero.
p>If Republicans would stop preventing broadband competition we'd be far better off. And before anyone wants to whine about being partisan, go take a look at the places which have outlawed municipal broadband. See the pattern?
California, Colorado, Minnesota, Nevada, Washington, Virginia all voted blue in the last presidential election, and all have some form of restriction or hurdle for municipal broadband. That's about 30% of the states with such regulations. The pattern is that of corporate corruption of politics, which affects both parties but Republicans more. Since you have solidly blue states such as CA and WA in on this, you really can't call it just a Republican problem.
My dream: local governments (or the local power company) run the "last mile" passive fiber to every home. Then any company can apply to come in and start hooking up at the switch boxes. This means new offerings like Google Fiber could hook up quickly, and the old guard can still provide competitive service if they choose to (also dragging them into a fiber-first model). No need to fret over who gets connected at the house level, because you have public oversight at that level, and not having to do the last-mile means there's less incentive to hook up only the rich neighborhoods, because all of them can be done fairly efficiently once you have backhaul. This design also keeps the government from trying to be an ISP, which they aren't really equipped for -- instead they maintain the street-level infrastructure, something they do a lot of already.
Before telecom deregulation I had a small ISP over Verizon's copper, and (for the time) it was great. The ISP of course got killed off as soon as Verizon was allowed to stop sharing the lines. A decade of stagnation followed. I'd love to see the smallest changes on the public side to make private competition viable, and a municipally owned last mile makes a lot of sense.
The mere $7K fine is perhaps the real travesty here. With a fine is that low it wouldn't even be in the company's financial interest to follow it, as the average cost of downtime DIVIDED by the probability of being caught is probably way more than $7K. The fine ought to be calculated explicitly to make sure it's cheaper to follow, that way even sociopathic management will want to do the right thing.
Ok, I have a PhD relevant to doomsday scenarios (robotics and AI, from a top university).
On the Bulletin's Science and Security Board, only 8/14 have PhDs, and most of those are related to environment or international policy. They don't have any scientists in the area of AI (overblown but nonzero threat) or biological warfare or disease (generally underestimated threat).
Will you listen to me?
The fact that Trump is a dipshit doesn't mean that the rest of us have to lose our minds. Many of my friends on social media have lost all ability to think or reason, and just pass through shoddy unsubstantiated articles as fact, which is sad because that's the problem with the POTUS that they are decrying in the first place. Fight idiocy with well sourced and reasoned explanations, and calm refusal to capitulate with the worst of it. Do not return in kind.
If you look at the Science and Security Board of the Bulletin, you'll see that only two of the 14 members have an atomic science background, and only two more have meaningful nuclear policy experience. This is very different from how it was when it started in 1945. Most of the people there now are environmental and public policy folks. Only 8/14 seem to have PhDs (what many would expect when you say "scientist").
It'd be more accurate to call it the Bulletin of Environmental Policy Scientists. In that lens their determination does make sense, as Trump will be nothing but bad for global warming. However a clock-to-midnight is a poor representation of a threat like that, which takes sustained and difficult work over a long period rather than a reduction of tensions to solve.
Apparently Elon Musk has tried to float the idea of a carbon tax with Trump. While unsuccessful so far, that's probably a bigger impact than the Bulletin will have during this administration.
The only "barriers" that exist are those created by Republican politicians.
Then explain why California has such a crappy broadband law. It does seem that deep Red states have the worst laws (outright bans), but the major ISPs are "friends" of every government and are just as happy with planting minefields. As long as it stops municipalities from solving the last mile, it's a good deal for them.
I don't think many people seriously believed voting itself was hacked.
Actually 50% of Clinton voters believe that, or about a quarter of the US population. Of course Trump voters had their own preferred conspiracy theories as you can see from the same report. What it shows is that people believe the narrative of their preferred news sources, and that neither major party is immune.
Rolling Stone has a nice article with an overview of the whole situation and why we should be skeptical. It's nice to see that coming from them, because MSNBC won't say it (gotta support out side) and nobody is going to believe Fox saying it, and sadly the rest of the press is starting to fall in line with one side or the other.
TFS should have included Google's response (already in TFA):
“We’ve worked hard to comply with the OFCCP’s current audit. However, the handful of OFCCP requests that are the subject of the complaint are overbroad in scope, or reveal confidential data, and we've made this clear to the OFCCP, to no avail. These requests include thousands of employees’ private contact information which we safeguard rigorously. We hope to continue working with OFCCP to resolve this matter.”
...reasonable organs come down the pipe.
I will never think of the word "organ pipe" quite the same way anymore.
Perhaps a better approach is Google's NaCl, where intermediate code is translated more directly to native code, while putting security guarantees in place. But here, of course, cross-platform support is an issue.
You might find Portable Native Client interesting, which is built out of LLVM+NaCl.
http://www.chromium.org/native...
I'm hoping that eventually browsers, mobile phones, and cloud hosting will just become sandboxed LLVM targets. Then people could use whatever language they like, wherever they want.
I live near Google and also work there. The bus stops are not anywhere near campus, as they are only intended for people with long commutes where pooling makes sense. I live 7 miles from my building, which is bikable and drivable, so no bus stop or bus for me. What you see around campus is empty buses during the day because they need somewhere to hang out between the morning and evening rush. I think you might be misinterpreting this as how they always look.
They are not driving them empty up and down the peninsula all day. Sometimes during rush hour there is one direction that'll be almost empty, but in those cases there are a *lot* of people the other way.
Many-to-one transportation systems (like corporate shuttles) are much easier to run efficiently than many-to-many systems (like most city buses). This is why systems use hubs -- even though they take you well out of your way, they are cheaper for the carrier. But in the case of a shuttle, the hub *is* where you want to go, so there is no loss of efficiency.
Manhattan has 1.6M people at an average density of 70,500 per sq mile. I don't know of any 1 Gbps service for $25 there.
How about where a company sinks several person-years of effort into a library or software suite, then decides to open source it? If an external developer spends a few days fixing a bug, it would be nice if the company that footed most of the development cost could use that fix in their internal deployments. Only fair, right?
Now hopefully it makes sense why companies release mostly BSD software (Clang being a good example). It is also why developers BSD things that they want companies to embrace (basic components where wide use helps like zlib, or things where you'd like a company to foot the long-term cost).
I've released both GPLv2 and MIT-licensed libraries. My current employer allows me to open source some work, but strongly encourages Apache licensing. I follow that, because they paid me to write it and it seems unfair to limit their future use.
Use quotes around words, or enable "verbatim results" in the options. The + thing was misunderstood and misused by most users, and they figured experts could RTFM.
The thesis you are basing your work is from 1977; while no doubt current when it was written, there is has been a lot of work on human signal decoding since then.
I'd strongly suggest looking at Hidden Markov Models:
http://en.wikipedia.org/wiki/Hidden_Markov_model
While some recent methods have gone beyond HMMs for speech recognition, that's been the baseline "good" solution for the past decade.
Since this is a binary signal problem another approach to consider would be Markov Random Fields (MRFs) which could be used as an initial de-noising pass or even as a full decoder if you set the cost functions right.
Your idea of user adaptation is pretty reasonable, but my guess is the primary thing that matters would be an overall speed scaling. IOW for good decoding you probably just need to normalize the average letter rate between users.
Good luck.
For public stop usage the SFMTA was aware of and already working toward a solution since late 2011:
https://www.sfmta.com/projects-planning/projects/shuttle-partners-program/detail
It's all laid out pretty reasonably without having to get into a ticketing war or protests. These protesters are late to this issue, yet will probably claim credit when the mutli-year regulation update goes in place next year.
Of course, this is just a side issue for the bus protesters, it is more about the evictions. There are a lot of things driving that from zoning regulations to economics, so they pick a visible if somewhat poorly representative target.
In your case, it does seems like quite a traffic growth problem, but replacing the each bus with 30+ cars doesn't seem like a good solution.