When I read the book a few years ago, the most striking thing was that the names Archer Daniels Midland and Dwayne Andreas didn't appear on the cover or liner notes, and I actually had to look pretty deep in the book to figure out what company he worked for.
Now that ADM pleaded guilty and paid a $100 million fine, their lawyers have less to work with, but at the time the publisher was apparently pretty scared of them. The book is also exceedingly careful about alleging that the Andreases knew how their company did business.
"Thanks. In fact, this is a very happy day for us - our team is top contender for winning the Grand Prize, as we have a better Test score than The Ensemble. (Probably this is the first post revealing this in the forum smile)"
I don't think this is quite clear. The teams submit their predictions on a qualifying set of about 1.8 million user/movie pairs. Half of these are used for the published leaderboard rankings, and the other half secretly scored for the actual prize. What the teams don't know is which user/movie ratings are in which half. The teams aren't submitting programs that Netflix runs though, they're submitting predictions on the whole quiz set.
I'm not a security expert, but the recent Clickjacking attacks and the older Cross-Site Request Forgery attacks seem to cast doubt on the wisdom of allowing one-click order execution in general. The recommended techniques for avoiding those attacks involve requiring extra deliberate user actions to confirm that the get/post wasn't generated by a browser hack.
Amazon might do enough testing to be relatively immune, but most smaller sites would probably want to avoid one-click orders to reduce the hacking risk, patent or not.
Yes, because it requires you to keep track of only a finite number of states to know everything you need to know about how you got to any given point in the sequence. That's the test for a regular language under the Myhill-Nerode theorem. However, writing the regex is going to be much more complicated than designing the state machine.
I worked out the regex for whether an integer is divisible by three a month or so ago and it's already pretty complicated although it works on the the same principle with fewer states. The Luhn one will be exponentially worse.
At the risk of doing your homework for you, the answer to your question is that according to the Myhill-Nerode theorem the language you describe is regular if and only if the inner pattern ("[FB][ot]o") matches only a finite number of strings. So yes in your example, but/(a+).*?\1/ is not strictly speaking a regular expression even though it's allowed in perl.
Interesting point, but it's probably no easier to replace the eprom that loads an FPGA than to make any other change to the device - routers have software already. And it wouldn't be an issue with an antifuse chip.
For some applications, it might be better to sacrifice performance and cost, and implement most of the logic in FPGAs. Then only one chip needs to be verified, and it can be bought from lots that were fabricated before you even finished your design.
If they think this approach is valuable to an enemy, what do you suppose the chances are that they aren't doing it themselves, but by pressuring the companies rather than surreptitiously inserting circuitry at the fab?
In the microprocessor case, suppose they added a bit of logic to look for a particular data sequence, and if found, switch to system management mode or ring 0 and execute whatever follows. Then they could take over any machine simply by sending it a data packet. Presumably there would be some code signing to prevent anyone else from exploiting the backdoor.
Intel, Cisco, et al are involved in the Critical Infrastructure Protection program and undoubtedly have other high-level contacts with the national security apparatus. It seems obvious that the US is in a better position than anyone else to carry out this type of attack.
I think part of the fun he's having at their expense is that it will cost them literally thousands of dollars just to have their lawyers read and digest that response - which they won't be able to avoid since they already authorized their lawyers to handle the communication with him.
1. Your personal salary is supposed to grow faster than the industry average salary. You advance in your career and gain experience, which gives you gains on top of the cited increase in pay for any particular job.
2. Compensation in tech is highly distorted relative to other professions. When Kai-Fu Lee got a huge ($10M) compensation package to leave MS for Google, his base pay was just $250,000. How newsworthy is it when a doctor or lawyer gets that?
Non-management engineering salaries pretty much top out at $140,000. Companies are using off-balance-sheet stock options to keep reported compensation costs low and report higher profits to investors. The problem is that the link between merit and income is much weaker in tech than in comparable professions. The result of the stock option lottery has been that way too many unimpressive people who were just in the right place at the right time have been bidding Bay Area houses beyond all reason. Top engineers are orders of magnitude more valuable than bad ones, but are lucky to be 50% better paid.
We tried using clickthroughs, but the data is very noisy. Users often don't know if a page is useful until they go to it, and they often open many pages from the same list of results. The best application turned out to be "how often is this link the last one people click on", but that assumes they're using the back button rater than opening several links in tabs.
You also don't know if the user finds what they really want linked off of a result page, or if they give up. The skewing of clicks toward the top results is very strong, but it seems to vary by the type of query. To even get this data, you need redirect links, and the fact Google doesn't use them tells me they didn't find it useful either.
There are two main directions where search can improve. One is better understanding of natural language, to disambiguate query terms and provide results where the wording on pages is different from the wording of the query.
The other, which this approach can address, is to improve the term relevance scores and overall page quality metrics that mainstream search engines are based on. Google had its initial success because of two features of this type: one was Page Rank, a measure of overall topic-independent site popularity, and two, bettor use of anchor text, the words people write when linking to other pages.
In both cases, they mined the link structure of the web, which was essentially aggregate community generated information about site quality that wasn't being spammed at the time. As they succeeded, regular people put less effort into writing their own link text, and spammers took over.
The next source of this type of community generated content will probably be something incidental instead of deliberately created. If you build a central repository of reviews of web sites, you both make it easy for people to game your results, and you open yourself up to lawsuits from interested parties.
However, untapped information already exists on what people find useful on the web in the form of their browsing histories, a special case of this being their bookmarks. Someone who could aggregate this information on what millions of people ended up looking at after they ran a particular search query would be in an excellent position to improve the traditional search engine scoring algorithm beyond link data.
I did a clean install on an x86-64 machine. I don't remember being prompted about whether to use the latest repositories. Yes, it installed and rebooted, but then I couldn't use the Add tool or yum to install any extra software off the DVD. This was after I'd had to resolve another problem where it didn't recognize my monitor correctly and put me in a broken video mode, so I was running out of patience. I did find some internet confirmation that there was a problem in FC6 with the.repo files, but it might just be in the 64bit version.
This shouldn't be underestimated as a legitimate reason.
In 2001, I needed this functionality, but I had an impossible time finding any documentation on whether it was even possible to do anything like XMLHttpRequest in a browser, especially in a cross platform way. One of the biggest changes is that we know what to search for now that the AJAX name took off following Google Maps' success.
As a C/Java programmer who had to cross over to some minor web development for an in-house app, I couldn't find the right search keywords to figure out how to do an unprompted request from the server. Maybe that's because the Microsoft execs didn't want to promote this guy's work.
Thank you, I was just about to post asking about this. I seem to remember not being able to update the bios with a bootable CD, but it probably depends on the motherboard.
I used --disablerepo to get rid of the ones that pointed to the web, and only included my custom DVD repo, but yum dies with another error about a media uri that didn't appear anywhere in the files. It mangled the file:///media/disk uri and put a bunch of random digits in front of it, then complained it doesn't exist. Maybe I could debug it if I actually had the relevant packages on the machine, but there's no reason to put up with this. I shouldn't even have to know about yum, let alone rpm, to install a fresh system on standard hardware. There's probably some magic answer I don't know about, but we should be past that by now.
I just installed FC6 on a machine yesterday, and they made it impossible to do anything without connecting to their server. I'm keeping the machine off the network, but apparently there's no way to install packages from the DVD without first downloading the update lists from their mirrors.
The Add/Remove gui (and yum) crashes if DNS isn't available. After some research, I was able to hack the yum.repo files to point to the DVD instead of the internet, but it still crashes with mysterious errors about media uris. I finally gave up and installed Ubuntu instead. So no, this doesn't help the whole Linux community. We'd be furious is Microsoft imposed this sort of requirement on new installations.
Call the VP or Human Resources and tell him what happened. Chances are he'll be furious that this is happening and straighten out the recruiter or whomever gave you the wrong information. He'll probably also be able to arrange an exception to give you what you expected, or at least a compromise.
No large company would do this as a matter of policy. No one wants employees who feel cheated; they don't work hard and might steal from the company.
What makes you think universities can change people that much? People who are going to be thinkers aren't going to be ruined by learning specific technologies. People who can't think creatively will find ways to learn by rote no matter what you test them on.
The problem is that the telcos are usually regional monopolies, and you absolutely have to go through certain backbone providers to get traffic to a given location.
Utilities aspire to be tax collectors, and without neutrality, they'll have little incentive to upgrade the speed of their free network.
Slashdot Mirror
When I read the book a few years ago, the most striking thing was that the names Archer Daniels Midland and Dwayne Andreas didn't appear on the cover or liner notes, and I actually had to look pretty deep in the book to figure out what company he worked for.
Now that ADM pleaded guilty and paid a $100 million fine, their lawyers have less to work with, but at the time the publisher was apparently pretty scared of them. The book is also exceedingly careful about alleging that the Andreases knew how their company did business.
http://www.netflixprize.com/community/viewtopic.php?id=1498
"Thanks. In fact, this is a very happy day for us - our team is top contender for winning the Grand Prize, as we have a better Test score than The Ensemble. (Probably this is the first post revealing this in the forum smile)"
Also, Yehuda Koren is at Yahoo now, not AT&T.
I don't think this is quite clear. The teams submit their predictions on a qualifying set of about 1.8 million user/movie pairs. Half of these are used for the published leaderboard rankings, and the other half secretly scored for the actual prize. What the teams don't know is which user/movie ratings are in which half. The teams aren't submitting programs that Netflix runs though, they're submitting predictions on the whole quiz set.
In fact, according to the second post by Yehuda Koren in this thread, it looks like BelKor does have the best test error rate and will be declared the winner. http://www.netflixprize.com/community/viewtopic.php?id=1498
I'm not a security expert, but the recent Clickjacking attacks and the older Cross-Site Request Forgery attacks seem to cast doubt on the wisdom of allowing one-click order execution in general. The recommended techniques for avoiding those attacks involve requiring extra deliberate user actions to confirm that the get/post wasn't generated by a browser hack.
Amazon might do enough testing to be relatively immune, but most smaller sites would probably want to avoid one-click orders to reduce the hacking risk, patent or not.
Yes, because it requires you to keep track of only a finite number of states to know everything you need to know about how you got to any given point in the sequence. That's the test for a regular language under the Myhill-Nerode theorem. However, writing the regex is going to be much more complicated than designing the state machine. I worked out the regex for whether an integer is divisible by three a month or so ago and it's already pretty complicated although it works on the the same principle with fewer states. The Luhn one will be exponentially worse.
At the risk of doing your homework for you, the answer to your question is that according to the Myhill-Nerode theorem the language you describe is regular if and only if the inner pattern ("[FB][ot]o") matches only a finite number of strings. So yes in your example, but /(a+).*?\1/ is not strictly speaking a regular expression even though it's allowed in perl.
Interesting point, but it's probably no easier to replace the eprom that loads an FPGA than to make any other change to the device - routers have software already. And it wouldn't be an issue with an antifuse chip.
For some applications, it might be better to sacrifice performance and cost, and implement most of the logic in FPGAs. Then only one chip needs to be verified, and it can be bought from lots that were fabricated before you even finished your design.
If they think this approach is valuable to an enemy, what do you suppose the chances are that they aren't doing it themselves, but by pressuring the companies rather than surreptitiously inserting circuitry at the fab?
In the microprocessor case, suppose they added a bit of logic to look for a particular data sequence, and if found, switch to system management mode or ring 0 and execute whatever follows. Then they could take over any machine simply by sending it a data packet. Presumably there would be some code signing to prevent anyone else from exploiting the backdoor.
Intel, Cisco, et al are involved in the Critical Infrastructure Protection program and undoubtedly have other high-level contacts with the national security apparatus. It seems obvious that the US is in a better position than anyone else to carry out this type of attack.
I think part of the fun he's having at their expense is that it will cost them literally thousands of dollars just to have their lawyers read and digest that response - which they won't be able to avoid since they already authorized their lawyers to handle the communication with him.
My first app would be AdBlock for real life.
Two points:
1. Your personal salary is supposed to grow faster than the industry average salary. You advance in your career and gain experience, which gives you gains on top of the cited increase in pay for any particular job.
2. Compensation in tech is highly distorted relative to other professions. When Kai-Fu Lee got a huge ($10M) compensation package to leave MS for Google, his base pay was just $250,000. How newsworthy is it when a doctor or lawyer gets that?
Non-management engineering salaries pretty much top out at $140,000. Companies are using off-balance-sheet stock options to keep reported compensation costs low and report higher profits to investors. The problem is that the link between merit and income is much weaker in tech than in comparable professions. The result of the stock option lottery has been that way too many unimpressive people who were just in the right place at the right time have been bidding Bay Area houses beyond all reason. Top engineers are orders of magnitude more valuable than bad ones, but are lucky to be 50% better paid.
The merchant who accepts the fraudulent charge eats the chargeback, not the one whose site is hacked. How does this encourage information security?
We tried using clickthroughs, but the data is very noisy. Users often don't know if a page is useful until they go to it, and they often open many pages from the same list of results. The best application turned out to be "how often is this link the last one people click on", but that assumes they're using the back button rater than opening several links in tabs.
You also don't know if the user finds what they really want linked off of a result page, or if they give up. The skewing of clicks toward the top results is very strong, but it seems to vary by the type of query. To even get this data, you need redirect links, and the fact Google doesn't use them tells me they didn't find it useful either.
There are two main directions where search can improve. One is better understanding of natural language, to disambiguate query terms and provide results where the wording on pages is different from the wording of the query.
The other, which this approach can address, is to improve the term relevance scores and overall page quality metrics that mainstream search engines are based on. Google had its initial success because of two features of this type: one was Page Rank, a measure of overall topic-independent site popularity, and two, bettor use of anchor text, the words people write when linking to other pages.
In both cases, they mined the link structure of the web, which was essentially aggregate community generated information about site quality that wasn't being spammed at the time. As they succeeded, regular people put less effort into writing their own link text, and spammers took over.
The next source of this type of community generated content will probably be something incidental instead of deliberately created. If you build a central repository of reviews of web sites, you both make it easy for people to game your results, and you open yourself up to lawsuits from interested parties.
However, untapped information already exists on what people find useful on the web in the form of their browsing histories, a special case of this being their bookmarks. Someone who could aggregate this information on what millions of people ended up looking at after they ran a particular search query would be in an excellent position to improve the traditional search engine scoring algorithm beyond link data.
I did a clean install on an x86-64 machine. I don't remember being prompted about whether to use the latest repositories. Yes, it installed and rebooted, but then I couldn't use the Add tool or yum to install any extra software off the DVD. This was after I'd had to resolve another problem where it didn't recognize my monitor correctly and put me in a broken video mode, so I was running out of patience. I did find some internet confirmation that there was a problem in FC6 with the .repo files, but it might just be in the 64bit version.
This shouldn't be underestimated as a legitimate reason.
In 2001, I needed this functionality, but I had an impossible time finding any documentation on whether it was even possible to do anything like XMLHttpRequest in a browser, especially in a cross platform way. One of the biggest changes is that we know what to search for now that the AJAX name took off following Google Maps' success.
As a C/Java programmer who had to cross over to some minor web development for an in-house app, I couldn't find the right search keywords to figure out how to do an unprompted request from the server. Maybe that's because the Microsoft execs didn't want to promote this guy's work.
Thank you, I was just about to post asking about this. I seem to remember not being able to update the bios with a bootable CD, but it probably depends on the motherboard.
I used --disablerepo to get rid of the ones that pointed to the web, and only included my custom DVD repo, but yum dies with another error about a media uri that didn't appear anywhere in the files. It mangled the file:///media/disk uri and put a bunch of random digits in front of it, then complained it doesn't exist. Maybe I could debug it if I actually had the relevant packages on the machine, but there's no reason to put up with this. I shouldn't even have to know about yum, let alone rpm, to install a fresh system on standard hardware. There's probably some magic answer I don't know about, but we should be past that by now.
I just installed FC6 on a machine yesterday, and they made it impossible to do anything without connecting to their server. I'm keeping the machine off the network, but apparently there's no way to install packages from the DVD without first downloading the update lists from their mirrors.
.repo files to point to the DVD instead of the internet, but it still crashes with mysterious errors about media uris. I finally gave up and installed Ubuntu instead. So no, this doesn't help the whole Linux community. We'd be furious is Microsoft imposed this sort of requirement on new installations.
The Add/Remove gui (and yum) crashes if DNS isn't available. After some research, I was able to hack the yum
Call the VP or Human Resources and tell him what happened. Chances are he'll be furious that this is happening and straighten out the recruiter or whomever gave you the wrong information. He'll probably also be able to arrange an exception to give you what you expected, or at least a compromise. No large company would do this as a matter of policy. No one wants employees who feel cheated; they don't work hard and might steal from the company.
What makes you think universities can change people that much? People who are going to be thinkers aren't going to be ruined by learning specific technologies. People who can't think creatively will find ways to learn by rote no matter what you test them on.
The problem is that the telcos are usually regional monopolies, and you absolutely have to go through certain backbone providers to get traffic to a given location. Utilities aspire to be tax collectors, and without neutrality, they'll have little incentive to upgrade the speed of their free network.
Great, now we have to post this every 108 minutes.