South Florida, definitely. It's amazing how many scams come from South Florida. There are whole classes of fraud from there not seen much elsewhere. Timeshare-sales fraud, phony DMV fraud, and phony tax-preparer fraud are examples.
In terms of dollar volume, though, lower Manhattan is way ahead. The South Floriday operators tend to be rather low rent.
Stanford has an elaborate VR lab. The system is 120FPS, and the lag is low, but I'm not sure how low. There's full motion tracking of the subject in a 20 foot by 20 foot space. They have public tours every Friday. Sign up and try high-end VR.
This isn't a graphics lab. It's a psychology lab. Some of the results are scary. They've had kids go through a VR experience of swimming with sharks. A few weeks later, the kids are asked about it, and a sizable fraction of them believe they really did it, adding details that were not in the sim like what they ate while visiting the sharks.
They're always running psychology experiments, and looking for volunteers. Pays $15/hr.
A college education is the single greatest value in the modern world. Nothing else even comes close. Dollar-for-dollar, nothing else delivers more quality of life to the individual -- nothing. People who complain about its cost have no idea what they are talking about.
That's just not true any more in the US. It's true for the top 10 colleges, and maybe the top 50, but as you go further down, the return on investment of a college degree goes negative now. In 2011, 85% of college students moved back in with their parents.
You're pretty much right. This type of process produces separate N2 and CO2 streams, and very little NOx. That's useful; the N2 is harmless, oxides of nitrogen are pollutants, and there are uses for reasonably pure CO2.
Well, that's the end of the Cell architecture. So much for a "build it and they will come" architecture. The Cell needed a theoretical breakthrough in software architecture to make it work, and that didn't happen.
The PS3 was the only non-shared-memory multiprocessor machine ever sold in volume. Supercomputers have been built with similar architectures, but in small quantities. Nobody ever tried mass marketing something like this before. Such architectures are notoriously hard to program. Having only 256K per CPU didn't help. Sony hoped that someone would figure out how to make it work. Most of Sony's software research staff at SCEA was diverted to trying to figure out how to program the PS3. They achieved some success, but it was much harder to program than the XBox So PS3 games were late and Sony lost market share. So Sony has admitted defeat, and produced what's basically an x86 PC in a set top box form factor.
The Cell might have worked if it had, say, 16MB per CPU. With only 256K, you couldn't render a frame, run a physics simulation, or do much interesting without having to DMA stuff in and out of the big memory. You had to rewrite everything to work by sequential streaming, like a DSP. The audio people loved it, because their problems work that way. Everybody else had to turn their algorithms inside out to cram them into the PS3. Or focus on a competing platform.
For three days, there's been flacking for Canonical's attempt, yet again, to commercialize Linux on end user devices. Canonical previously claimed their product was going to ship on the EeePC (it never reached retail channels), and on Dell (where it was more expensive than Windows).
Worse, Canonical ships a Linux preloaded with ad-supported crapware.
Here's a typical Google-hosted phishing page. Note that the page is long enough that the Google disclaimers at the bottom are pushed "below the fold", and some users won't notice. Such pages are used in conjunction with spam emails. Since the URL in the spam will be on Google, it makes it through most spam filters.
Google's own phishing detection catches some of these. Ones that mention "Microsoft Outlook" tend to be caught. This suggests that Google is using a simple classifier but needs a better training set. There's enough similarity between most of the fake login pages that many are clearly coming from the same sources or the same toolkits. It looks like there are only about two or three different attackers exploiting Google, and they're not working very hard at making convincing fake login pages. Or maybe the better-funded attacks aren't being detected by this approach.
One of the things our SiteTruth system does is report on major sites that host phishing scams. There are only 34 such sites today. As it has been for several years now, Google is at the top of the list.
Here's the list of
all known phishing sites currently hosted by Google..
Scroll down through all that background data about the company to a big block of red "phishtank report (2013-02-01): Phony site reported via PhishTank." lines. Click on the links for a PhishTank report.
The raw data comes mostly from PhishTank. Most exploitable hosting services (especially short-URL services) check PhishTank and the APWG list automatically, but not Google.
Google has several vulnerabilities. It's possible to host an attack page not only on Google Sites and Google Docs, but also on Google Spreadsheets. Recently, Google added a new attack vector; there's an open redirector at Google Accounts.
Amusingly, for some, but not all, of these phishing sites, Google's own anti-phishing warning pops up. But the part of Google that generates that blacklist clearly doesn't talk to the part of Google that does hosting.
Here's the oldest phishing site hosted by Google. On line since 2010-12-30. It's one of those "Habbo Coins" phishing pages, probably forgotten by the original attacker, since it forwards to a dead Hotmail account.
When we first started doing this analysis, Google wasn't on the list, because they didn't do hosting. There were about 150 sites listed in 2009. Through improved awareness, nagging and the Anti-Phishing Working Group, we're down to 34 - a few little sites with no clue, ones that just got hit by break-ins, and "bit.ly", which tries to keep up with their abuse problem but is falling behind. MSN, Yahoo, TinyURL, and most of the other big-time victims long ago solved their problems in this area. Google stands alone as a major service with an incompetent abuse department.
CALTRANS had an induction-charged bus deployed in Berkeley in the 1980s. It required precise parking at bus stops, so the two halves of the split transformer could connect magnetically. The system worked OK, but wasn't a huge win.
GE once patented a system where an entire lane had transformers, so vehicles could run on ground power. That was too expensive. It would cost like a maglev track.
This sounds like a pork program. When I read about computer companies talking to the Government about there not being "enough compute resources", I think of the various supercomputer boondoggles. Here's the current job list for the Mississippi Supercomputer Center. Look at the CPU and memory usage columns. Most, if not all, of those jobs could be running on a 4-core 64-bit desktop machine. Instead, they're running some 10-year old SGI supercomputers as a batch processing service, free to Mississippi academics.
Speaking as an inventor with six patents, it's hard to enforce a patent until you have $100 million in infringing activity. In practice, almost everybody who gets a royalty deal gets about 5%, +- 2%. (There's a whole theory of IP valuation, but it's not taken very seriously.) So 5% of $100 million is $5 million. Expect legal fees of about $2 million. So you make about $3 million, best case. It's fully taxable, so you get to keep about $2 million. The odds of winning a patent case are about 30%-40%, So the expectation is about $700K on $100,000,000 in infringement.
I've licensed two patents. One I swapped for stock in a startup, and that came out very well. There I wrote one of the startup's products. A straight licensing deal on another made me about $400K after taxes; that was partly about getting my product off the market so it didn't compete with theirs. I'm working on licensing the other patents.
Radios with a generator charger aren't new. I had one in the 1980s, and they go back at least to WWII. The problem was that no battery charging occurs unless you're cranking hard enough to get the generator voltage above the battery voltage. Then you have to have a governor or voltage limiter to prevent overvoltaging the battery. So you have to crank at just the right speed, and there has to be a mechanism to enforce this.
So this guy used a clockwork mechanism with a big spring to power a small generator, which was more convenient. You can wind fast, or you can wind slow; it doesn't matter. Then it occurred to others that DC voltage conversion isn't hard any more, and it's straightforward to build a charger that will work off whatever is coming from the generator. So you get rid of the clockwork and replace it with a crank generator, switching power supply and battery. The OLPC has a gadget like that.
Most free "cloud" services only seem to last a few years, until the vendor realizes they're not making money. Using a big-name vendor doesn't help - remember Google Wave, Apple MobileMe, Wal-Mart Music, Microsoft Windows Live, etc. The lives of these things are surprisingly short. About as long as a cool restaurant.
Don't get locked into a "cloud" service that
stores your data in a form that can't be readily exported to somewhere else.
At some point China will probably do a moon landing. It won't benefit them any more than it did the US, but it will be so embarrassing. Especially if they send the remains of the US flag back.
I'm not sure I see a problem here. The seller is told who the buyer is. That's reasonable enough. It also keeps Google honest with respect to sellers - you can have some people make test buys and make sure that Google pays you for them.
I'm generally critical of Google's non-approach to non-privacy, but here, there's a real transaction, with money.
What are the other alternatives? Over the years, Visual Basic, and Borland Delphi were tried. They're now half-dead languages. Stored procedures in databases and various XML-based hacks were used in some firms. Almost all these solutions are single vendor, which means you get overcharged or abandoned. Sometimes both.
I don't think many on Rent-A-Coder can actually program. I once put a job on there. I had a simple Python program that would retrieve and format in a standard form WHOIS data from one registrar's WHOIS server. I wanted modules written for about 50 other registrars. That's a simple formatting job; I just didn't want to write all the variants. Three Rent-A-Coder "programmers" in succession tried and gave up. Not one ever delivered a single line of code. This wasn't exactly rocket science.
I tried "freelancer.com" once for some simple web design work. I was willing to pay about $500 for one well-designed page with some specific original artwork. I got back crap clip art. I finally paid $500 to a student at the California College of the Arts in San Francisco, and got back good work in a week with no problems.
"Freelancer.com" was difficult about returning my money. I discovered that the regional small claims court in Australia accepts online filings. I filled out the appropriate online forms, paid a small court fee, and within hours of filing a case, Freelancer sent me a refund by wire transfer.
While, in the US (or even the EU), we're not likely to see a "right to be forgotten", we might have a "right not to have one's identity exploited for advertising purposes". You should be able to quit an ad-supported service and insist that none of your data every appear on a page with an ad. If it does, the advertiser has to pay you a publicity fee. California has a law like that for photos - if you use someone's photo in an ad without their permission, you owe them at least $500 - much more if they're famous.
The nice thing is the way the system grades our answers - it actually evaluates our answers to see if they are mathematically equal to the desired answer (within certain limits, for example if it asks me to integrate something, I can't just put in the integral, I actually have to do the work) so, basically if the answer is Pi/3, it will also accept 3Pi/9, etc.
Plato had that feature 40 years ago. It checked to see if two equations were equivalent by generating random test cases for the variables and checking for near numeric equivalence.
There are far too many "web developers" already. Web development is an additional skill for copywriters, illustrators, and office admins. In 1998, you could make big money as a "web developer". Back then, it was important to get into the field fast and not miss out on the dot-com boom. That's over.
Most web development today uses some "framework" or content management system. Users of those need only a modest level of training. Weeks, not years. It's more like learning Microsoft Office.
What happens when leftist developers don't rip off UNIX, Windows or MacOSX.
There's something to be said for this. There are things that open source development does not do well. Designing tight, elegant systems from scratch is one of them. The Hurd crowd started hacking on Mach. That sucked. Then they tried L4. That didn't work. Then they tried Coyotos. That didn't work.
The QNX kernel is about 60K bytes of x86 code. Writing a microkernel is not about writing a lot of software. It's about writing a small amount of very well designed software. The hack and patch approach doesn't work for that.
I know. QNX does a lot of things right. The company, though, is notorious for driving its customers and employees nuts. It's been sold twice, one to Harmon (car audio) and then to RIM (now Blackberry). The code went from closed source to open source to closed source to open source to closed source. During the latter part of the Harmon period, you could download the entire kernel source.
The developer community was fed up by this. During the open source periods, there were QNX builds for many major open source products, like Firebird (what Firefox was first called) and GCC. Those are no longer maintained.
The QNX kernel is only about 60K bytes on x86 platforms. All it does is message passing, CPU dispatching, memory management, and timers. There's also a built-in process called "proc", which is a few hundred K. All device drivers, file systems, and networking are in user space. One of the great things about having such a tiny kernel is that it can be fully debugged. It needs to be changed very rarely. It can be put in ROM and stay unchanged for the life of the machine. In many embedded applications, it is. If the Hurd kernel is much bigger than that, they're doing it wrong.
You can still get QNX for free for non-commercial purposes. Few people do.
Slashdot Mirror
I think fraud is simply in Florida's DNA.
South Florida, definitely. It's amazing how many scams come from South Florida. There are whole classes of fraud from there not seen much elsewhere. Timeshare-sales fraud, phony DMV fraud, and phony tax-preparer fraud are examples.
In terms of dollar volume, though, lower Manhattan is way ahead. The South Floriday operators tend to be rather low rent.
Stanford has an elaborate VR lab. The system is 120FPS, and the lag is low, but I'm not sure how low. There's full motion tracking of the subject in a 20 foot by 20 foot space. They have public tours every Friday. Sign up and try high-end VR.
This isn't a graphics lab. It's a psychology lab. Some of the results are scary. They've had kids go through a VR experience of swimming with sharks. A few weeks later, the kids are asked about it, and a sizable fraction of them believe they really did it, adding details that were not in the sim like what they ate while visiting the sharks.
They're always running psychology experiments, and looking for volunteers. Pays $15/hr.
A college education is the single greatest value in the modern world. Nothing else even comes close. Dollar-for-dollar, nothing else delivers more quality of life to the individual -- nothing. People who complain about its cost have no idea what they are talking about.
That's just not true any more in the US. It's true for the top 10 colleges, and maybe the top 50, but as you go further down, the return on investment of a college degree goes negative now. In 2011, 85% of college students moved back in with their parents.
I could be wrong.
You're pretty much right. This type of process produces separate N2 and CO2 streams, and very little NOx. That's useful; the N2 is harmless, oxides of nitrogen are pollutants, and there are uses for reasonably pure CO2.
Amusingly, Google has now killed that page, presumably from embarrassment. But here's another one just like it, hosted by Google since 2010.
Well, that's the end of the Cell architecture. So much for a "build it and they will come" architecture. The Cell needed a theoretical breakthrough in software architecture to make it work, and that didn't happen.
The PS3 was the only non-shared-memory multiprocessor machine ever sold in volume. Supercomputers have been built with similar architectures, but in small quantities. Nobody ever tried mass marketing something like this before. Such architectures are notoriously hard to program. Having only 256K per CPU didn't help. Sony hoped that someone would figure out how to make it work. Most of Sony's software research staff at SCEA was diverted to trying to figure out how to program the PS3. They achieved some success, but it was much harder to program than the XBox So PS3 games were late and Sony lost market share. So Sony has admitted defeat, and produced what's basically an x86 PC in a set top box form factor.
The Cell might have worked if it had, say, 16MB per CPU. With only 256K, you couldn't render a frame, run a physics simulation, or do much interesting without having to DMA stuff in and out of the big memory. You had to rewrite everything to work by sequential streaming, like a DSP. The audio people loved it, because their problems work that way. Everybody else had to turn their algorithms inside out to cram them into the PS3. Or focus on a competing platform.
For three days, there's been flacking for Canonical's attempt, yet again, to commercialize Linux on end user devices. Canonical previously claimed their product was going to ship on the EeePC (it never reached retail channels), and on Dell (where it was more expensive than Windows). Worse, Canonical ships a Linux preloaded with ad-supported crapware.
Linux tablets are available from China. Some are good, some are awful, most are cheap.
New? That went in a few Firefox versions back, I think at Firefox 16. I turned it off, since I use Sumatra PDF (which is dumb, but safe).
Here's a typical Google-hosted phishing page. Note that the page is long enough that the Google disclaimers at the bottom are pushed "below the fold", and some users won't notice. Such pages are used in conjunction with spam emails. Since the URL in the spam will be on Google, it makes it through most spam filters.
Google's own phishing detection catches some of these. Ones that mention "Microsoft Outlook" tend to be caught. This suggests that Google is using a simple classifier but needs a better training set. There's enough similarity between most of the fake login pages that many are clearly coming from the same sources or the same toolkits. It looks like there are only about two or three different attackers exploiting Google, and they're not working very hard at making convincing fake login pages. Or maybe the better-funded attacks aren't being detected by this approach.
One of the things our SiteTruth system does is report on major sites that host phishing scams. There are only 34 such sites today. As it has been for several years now, Google is at the top of the list.
Here's the list of all known phishing sites currently hosted by Google.. Scroll down through all that background data about the company to a big block of red "phishtank report (2013-02-01): Phony site reported via PhishTank." lines. Click on the links for a PhishTank report. The raw data comes mostly from PhishTank. Most exploitable hosting services (especially short-URL services) check PhishTank and the APWG list automatically, but not Google.
Google has several vulnerabilities. It's possible to host an attack page not only on Google Sites and Google Docs, but also on Google Spreadsheets. Recently, Google added a new attack vector; there's an open redirector at Google Accounts.
Amusingly, for some, but not all, of these phishing sites, Google's own anti-phishing warning pops up. But the part of Google that generates that blacklist clearly doesn't talk to the part of Google that does hosting.
Here's the oldest phishing site hosted by Google. On line since 2010-12-30. It's one of those "Habbo Coins" phishing pages, probably forgotten by the original attacker, since it forwards to a dead Hotmail account.
When we first started doing this analysis, Google wasn't on the list, because they didn't do hosting. There were about 150 sites listed in 2009. Through improved awareness, nagging and the Anti-Phishing Working Group, we're down to 34 - a few little sites with no clue, ones that just got hit by break-ins, and "bit.ly", which tries to keep up with their abuse problem but is falling behind. MSN, Yahoo, TinyURL, and most of the other big-time victims long ago solved their problems in this area. Google stands alone as a major service with an incompetent abuse department.
CALTRANS had an induction-charged bus deployed in Berkeley in the 1980s. It required precise parking at bus stops, so the two halves of the split transformer could connect magnetically. The system worked OK, but wasn't a huge win.
GE once patented a system where an entire lane had transformers, so vehicles could run on ground power. That was too expensive. It would cost like a maglev track.
This sounds like a pork program. When I read about computer companies talking to the Government about there not being "enough compute resources", I think of the various supercomputer boondoggles. Here's the current job list for the Mississippi Supercomputer Center. Look at the CPU and memory usage columns. Most, if not all, of those jobs could be running on a 4-core 64-bit desktop machine. Instead, they're running some 10-year old SGI supercomputers as a batch processing service, free to Mississippi academics.
Speaking as an inventor with six patents, it's hard to enforce a patent until you have $100 million in infringing activity. In practice, almost everybody who gets a royalty deal gets about 5%, +- 2%. (There's a whole theory of IP valuation, but it's not taken very seriously.) So 5% of $100 million is $5 million. Expect legal fees of about $2 million. So you make about $3 million, best case. It's fully taxable, so you get to keep about $2 million. The odds of winning a patent case are about 30%-40%, So the expectation is about $700K on $100,000,000 in infringement.
I've licensed two patents. One I swapped for stock in a startup, and that came out very well. There I wrote one of the startup's products. A straight licensing deal on another made me about $400K after taxes; that was partly about getting my product off the market so it didn't compete with theirs. I'm working on licensing the other patents.
Radios with a generator charger aren't new. I had one in the 1980s, and they go back at least to WWII. The problem was that no battery charging occurs unless you're cranking hard enough to get the generator voltage above the battery voltage. Then you have to have a governor or voltage limiter to prevent overvoltaging the battery. So you have to crank at just the right speed, and there has to be a mechanism to enforce this.
So this guy used a clockwork mechanism with a big spring to power a small generator, which was more convenient. You can wind fast, or you can wind slow; it doesn't matter. Then it occurred to others that DC voltage conversion isn't hard any more, and it's straightforward to build a charger that will work off whatever is coming from the generator. So you get rid of the clockwork and replace it with a crank generator, switching power supply and battery. The OLPC has a gadget like that.
Nice. Reminds me of Speedtree's demos.
Most free "cloud" services only seem to last a few years, until the vendor realizes they're not making money. Using a big-name vendor doesn't help - remember Google Wave, Apple MobileMe, Wal-Mart Music, Microsoft Windows Live, etc. The lives of these things are surprisingly short. About as long as a cool restaurant.
Don't get locked into a "cloud" service that stores your data in a form that can't be readily exported to somewhere else.
At some point China will probably do a moon landing. It won't benefit them any more than it did the US, but it will be so embarrassing. Especially if they send the remains of the US flag back.
I'm not sure I see a problem here. The seller is told who the buyer is. That's reasonable enough. It also keeps Google honest with respect to sellers - you can have some people make test buys and make sure that Google pays you for them.
I'm generally critical of Google's non-approach to non-privacy, but here, there's a real transaction, with money.
Right. What's the alternative? Java? A language so insecure that the Department of Homeland Security told people to disable it. Where you have a huge pile of badly debugged libraries which keep changing.
What are the other alternatives? Over the years, Visual Basic, and Borland Delphi were tried. They're now half-dead languages. Stored procedures in databases and various XML-based hacks were used in some firms. Almost all these solutions are single vendor, which means you get overcharged or abandoned. Sometimes both.
I don't think many on Rent-A-Coder can actually program. I once put a job on there. I had a simple Python program that would retrieve and format in a standard form WHOIS data from one registrar's WHOIS server. I wanted modules written for about 50 other registrars. That's a simple formatting job; I just didn't want to write all the variants. Three Rent-A-Coder "programmers" in succession tried and gave up. Not one ever delivered a single line of code. This wasn't exactly rocket science.
I tried "freelancer.com" once for some simple web design work. I was willing to pay about $500 for one well-designed page with some specific original artwork. I got back crap clip art. I finally paid $500 to a student at the California College of the Arts in San Francisco, and got back good work in a week with no problems.
"Freelancer.com" was difficult about returning my money. I discovered that the regional small claims court in Australia accepts online filings. I filled out the appropriate online forms, paid a small court fee, and within hours of filing a case, Freelancer sent me a refund by wire transfer.
While, in the US (or even the EU), we're not likely to see a "right to be forgotten", we might have a "right not to have one's identity exploited for advertising purposes". You should be able to quit an ad-supported service and insist that none of your data every appear on a page with an ad. If it does, the advertiser has to pay you a publicity fee. California has a law like that for photos - if you use someone's photo in an ad without their permission, you owe them at least $500 - much more if they're famous.
The nice thing is the way the system grades our answers - it actually evaluates our answers to see if they are mathematically equal to the desired answer (within certain limits, for example if it asks me to integrate something, I can't just put in the integral, I actually have to do the work) so, basically if the answer is Pi/3, it will also accept 3Pi/9, etc.
Plato had that feature 40 years ago. It checked to see if two equations were equivalent by generating random test cases for the variables and checking for near numeric equivalence.
There are far too many "web developers" already. Web development is an additional skill for copywriters, illustrators, and office admins. In 1998, you could make big money as a "web developer". Back then, it was important to get into the field fast and not miss out on the dot-com boom. That's over.
Most web development today uses some "framework" or content management system. Users of those need only a modest level of training. Weeks, not years. It's more like learning Microsoft Office.
What happens when leftist developers don't rip off UNIX, Windows or MacOSX.
There's something to be said for this. There are things that open source development does not do well. Designing tight, elegant systems from scratch is one of them. The Hurd crowd started hacking on Mach. That sucked. Then they tried L4. That didn't work. Then they tried Coyotos. That didn't work.
The QNX kernel is about 60K bytes of x86 code. Writing a microkernel is not about writing a lot of software. It's about writing a small amount of very well designed software. The hack and patch approach doesn't work for that.
I know. QNX does a lot of things right. The company, though, is notorious for driving its customers and employees nuts. It's been sold twice, one to Harmon (car audio) and then to RIM (now Blackberry). The code went from closed source to open source to closed source to open source to closed source. During the latter part of the Harmon period, you could download the entire kernel source.
The developer community was fed up by this. During the open source periods, there were QNX builds for many major open source products, like Firebird (what Firefox was first called) and GCC. Those are no longer maintained.
The QNX kernel is only about 60K bytes on x86 platforms. All it does is message passing, CPU dispatching, memory management, and timers. There's also a built-in process called "proc", which is a few hundred K. All device drivers, file systems, and networking are in user space. One of the great things about having such a tiny kernel is that it can be fully debugged. It needs to be changed very rarely. It can be put in ROM and stay unchanged for the life of the machine. In many embedded applications, it is. If the Hurd kernel is much bigger than that, they're doing it wrong.
You can still get QNX for free for non-commercial purposes. Few people do.