Doesn't look like it at the moment. It's a new feature in glibc 2.10, which isn't yet in Debian (not even unstable, though there's a version in experimental) or in the latest Ubuntu, though it looks like it's in the dev versions of the upcoming late-October Ubuntu release.
Also, FWIW, pass-through income being taxed at both the pass-through point and the ultimate destination isn't unique to corporations. If I earn $1,000,000 this year, and then give $500,000 of it to my brother, I must pay taxes on the full $1,000,000, and my brother must pay taxes on the $500,000, so the $500k is double-taxed.
There are plenty of people who're taxed but not able to vote. Non-citizen permanent residents, those under 18, convicted felons, etc., all must still pay taxes. Do you propose exempting them all from taxes as well?
Because corporations are legal persons, so should pay taxes just like all other people do.
If you want to abolish corporate personhood, then sure, we can abolish corporate taxation too. But you can't count corporations as just proxies for individuals in one case, and not in another.
(That is, while in logic, a slippery slope argument is a kind of fallacy [they aren't logically inevitable], in the real world, many kinds of political change do in practice resemble a slippery slope, where each successive change makes it easier to introduce the next one.)
Hmm, I could see that, though the Atari VCS (aka Atari 2600) was a pretty popular platform. There's fan forums and such devoted to it. It seemed cool to me that someone finally wrote a good book on it, since it was pretty influential.
Upside: shows confidence in your products; makes it more likely that your engineers will spot problems if they use the software and services themselves; can increase how motivated people are to improve the products
Downside: tainted dogfood kills the engineers who would have investigated the issue
Yeah, it's true that there's some pretty lame stuff on the bioinformatics side too--- especially the early stuff has a feel of "hey guys what is computer", with books like Beginning Perl for Biologists.
Yeah, I probably should've been nicer. =] The Slashdot summary is actually more objectionable than the article is: as you point out, the metaphors in the article are quite well done. If you don't view it as "l33t XBox hacker discovers how to haxx0r viruses", but instead as "engaging tech writer uses computer terminology to explain how viruses work", it's much better.
That said, it's a quite well-written tutorial-style article with engaging prose that tackles a number of the relevant issues. I just balked at the "reverse engineer takes on biology" angle, as if that were something biologists had never thought of.
The SDK part is still free afaik, but it only works with Visual Studio, which isn't. Apple differs by distributing the whole toolchain free (iPhone SDK + XCode).
If only biologists had thought of the idea of treating DNA/RNA sequences as data, and then analyzing their properties statistically and computationally, with an eye towards what effects different modifications to the sequences might be predicted to have. We might call this field something fancy like "biological informatics".
Yeah, the term dates back at least to the 1990s. The classic survey paper (over 1000 citations!) on the subject is "Ensemble Methods in Machine Learning" [pdf] by Tom Dietterich (2000), for those who want to glance through a survey. Though be warned that some of its specific conclusions are now dated--- e.g. there's been a *lot* written in both statistics and machine learning since then on what boosting "really" is and why it works.
Dietterich presents the more machine-learning view of it, focused on algorithms, combination of predictions, iterative refinement, etc. The best survey from a statistical approach is probably Ch. 16 of this book by threeStanfordprofs, which you can probably read some of on Google Books.
Making a prediction by running multiple statistical prediction algorithms and combining their results often seems to work well. This is called an "ensemble method". Ensemble methods seem to work particularly well on collaborative filtering problems.
Although that's true with humans, it's a bit curious why it'd be true with algorithms. After all, the aggregation of 3 algorithms is still just an algorithm. It's not even totally clear which algorithms are ensembles and which aren't--- some non-ensemble methods could be re-analyzed using ensemble terminology, and some ensemble methods could be rewritten as unified iterative loops that don't look very ensemble-y. The jury's still out on the whole subject, as far as I can tell (I'm not an ML person, but I'm an AI person whose research bleeds into ML).
An exception is when you're aggregating information from truly different statistical problems, in which case you inherently have an ensemble problem, until someone comes up with the theory (plus tractable implementation) to view the problem as one unified statistical problem. I think collaborative filtering is currently in that stage--- there's no canonical way to pose the problem in the terminology of statistical regression/etc. that captures all aspects of it.
There's a lot of argument over why ensemble techniques work well in general, when using them on well-posed statistical problems. But in the collaborative filtering case, they work well at least in part because there's not a canonical way of posing the problem statistically that's also tractable--- there are instead multiple ways to view the problem, which expose different information. Aggregating those views is a pretty straightforward way of getting more information.
For example, you can see the Netflix prize as a few different standard statistical problems. As a per-movie regression, predicting what Person A will rate Movie B, given ratings vector of Person A and the ratings vectors of everyone who's already rated Movie B [the per-person ratings vectors excluding B are the X's, and the ratings on B are the Y's]. Or you slice the movie-ratings matrix the other way, with per-movie ratings vectors as the X's. Add in some other views (those are the two most straightforward), aggregate all the info you get from them, and you do better than any one approach alone.
Aren't devices improvised in this manner a pretty standard part of warfare, though? Digging tunnels and filling them with miscellaneous explosives, which you then try to detonate under (or at least near) your enemy, has been done for hundreds of years, and nobody calls e.g. Messines an IED attack.
I'm not a big fan of the GOP but I never heard a GOP Senator suggest withholding highway funds to blackmail the states into outlawing texting while driving.
I suppose it's been a while, but what about the Reagan administration's National Minimum Drinking Age Act, which threatened withholding highway funds to blackmail the states into raising their drinking ages from 18 to 21?
Actually Jefferson likely would have been appalled by the fact that the Federal Government is so heavily involved in education.
I agree Jefferson was much more in favor of state systems (George Washington, to the contrary, was in favor of a national education system). But that's still not really libertarian, which would argue we shouldn't have universal free public education at all. On the state level, Jefferson pushed tirelessly for an expansive system of free public schools in Virginia (mostly unsuccessfully), and was heavily involved in setting up the state-run University of Virginia.
And I'm not sure quite how solid Jefferson's opposition to federal involvement was: he actually proposed a constitutional amendment in 1806 that would place education "among the articles of public care" so that the federal government could "come in aid of the public education".
Re:Early detection doesn't always improve outcomes
on
A Breathalyzer For Cancer
·
· Score: 3, Insightful
It depends on how you value various things. Detecting a lot of people who would not have died anyway in Phase 1 will raise your overall survival rates (those people will still live), but may make a whole lot of people's lives shittier. Is removing an X% change you'd have died worth a Y% chance that you unnecessarily made your life shitty for years? Depends on X, Y, and your preferences.
Slashdot Mirror
Doesn't look like it at the moment. It's a new feature in glibc 2.10, which isn't yet in Debian (not even unstable, though there's a version in experimental) or in the latest Ubuntu, though it looks like it's in the dev versions of the upcoming late-October Ubuntu release.
Also, FWIW, pass-through income being taxed at both the pass-through point and the ultimate destination isn't unique to corporations. If I earn $1,000,000 this year, and then give $500,000 of it to my brother, I must pay taxes on the full $1,000,000, and my brother must pay taxes on the $500,000, so the $500k is double-taxed.
There are plenty of people who're taxed but not able to vote. Non-citizen permanent residents, those under 18, convicted felons, etc., all must still pay taxes. Do you propose exempting them all from taxes as well?
Because corporations are legal persons, so should pay taxes just like all other people do.
If you want to abolish corporate personhood, then sure, we can abolish corporate taxation too. But you can't count corporations as just proxies for individuals in one case, and not in another.
The slippery slope really does exist!
(That is, while in logic, a slippery slope argument is a kind of fallacy [they aren't logically inevitable], in the real world, many kinds of political change do in practice resemble a slippery slope, where each successive change makes it easier to introduce the next one.)
Indeed, the proper English second-person plural pronoun is "yinz".
Hmm, I could see that, though the Atari VCS (aka Atari 2600) was a pretty popular platform. There's fan forums and such devoted to it. It seemed cool to me that someone finally wrote a good book on it, since it was pretty influential.
What's that cryptic comment mean? I did indeed buy it, though, yes. =]
Upside: shows confidence in your products; makes it more likely that your engineers will spot problems if they use the software and services themselves; can increase how motivated people are to improve the products
Downside: tainted dogfood kills the engineers who would have investigated the issue
Yeah, it's true that there's some pretty lame stuff on the bioinformatics side too--- especially the early stuff has a feel of "hey guys what is computer", with books like Beginning Perl for Biologists.
Yeah, I probably should've been nicer. =] The Slashdot summary is actually more objectionable than the article is: as you point out, the metaphors in the article are quite well done. If you don't view it as "l33t XBox hacker discovers how to haxx0r viruses", but instead as "engaging tech writer uses computer terminology to explain how viruses work", it's much better.
(Replying to my own comment.)
That said, it's a quite well-written tutorial-style article with engaging prose that tackles a number of the relevant issues. I just balked at the "reverse engineer takes on biology" angle, as if that were something biologists had never thought of.
The SDK part is still free afaik, but it only works with Visual Studio, which isn't. Apple differs by distributing the whole toolchain free (iPhone SDK + XCode).
If only biologists had thought of the idea of treating DNA/RNA sequences as data, and then analyzing their properties statistically and computationally, with an eye towards what effects different modifications to the sequences might be predicted to have. We might call this field something fancy like "biological informatics".
Yeah, the term dates back at least to the 1990s. The classic survey paper (over 1000 citations!) on the subject is "Ensemble Methods in Machine Learning" [pdf] by Tom Dietterich (2000), for those who want to glance through a survey. Though be warned that some of its specific conclusions are now dated--- e.g. there's been a *lot* written in both statistics and machine learning since then on what boosting "really" is and why it works.
Dietterich presents the more machine-learning view of it, focused on algorithms, combination of predictions, iterative refinement, etc. The best survey from a statistical approach is probably Ch. 16 of this book by three Stanford profs, which you can probably read some of on Google Books.
Here's a stab at 3 sentences:
Making a prediction by running multiple statistical prediction algorithms and combining their results often seems to work well. This is called an "ensemble method". Ensemble methods seem to work particularly well on collaborative filtering problems.
Although that's true with humans, it's a bit curious why it'd be true with algorithms. After all, the aggregation of 3 algorithms is still just an algorithm. It's not even totally clear which algorithms are ensembles and which aren't--- some non-ensemble methods could be re-analyzed using ensemble terminology, and some ensemble methods could be rewritten as unified iterative loops that don't look very ensemble-y. The jury's still out on the whole subject, as far as I can tell (I'm not an ML person, but I'm an AI person whose research bleeds into ML).
An exception is when you're aggregating information from truly different statistical problems, in which case you inherently have an ensemble problem, until someone comes up with the theory (plus tractable implementation) to view the problem as one unified statistical problem. I think collaborative filtering is currently in that stage--- there's no canonical way to pose the problem in the terminology of statistical regression/etc. that captures all aspects of it.
There's a lot of argument over why ensemble techniques work well in general, when using them on well-posed statistical problems. But in the collaborative filtering case, they work well at least in part because there's not a canonical way of posing the problem statistically that's also tractable--- there are instead multiple ways to view the problem, which expose different information. Aggregating those views is a pretty straightforward way of getting more information.
For example, you can see the Netflix prize as a few different standard statistical problems. As a per-movie regression, predicting what Person A will rate Movie B, given ratings vector of Person A and the ratings vectors of everyone who's already rated Movie B [the per-person ratings vectors excluding B are the X's, and the ratings on B are the Y's]. Or you slice the movie-ratings matrix the other way, with per-movie ratings vectors as the X's. Add in some other views (those are the two most straightforward), aggregate all the info you get from them, and you do better than any one approach alone.
You're lucky I don't have my Light Gun handy.
You sure you aren't a WashU recruiter? ;-)
Aren't devices improvised in this manner a pretty standard part of warfare, though? Digging tunnels and filling them with miscellaneous explosives, which you then try to detonate under (or at least near) your enemy, has been done for hundreds of years, and nobody calls e.g. Messines an IED attack.
I suppose it's been a while, but what about the Reagan administration's National Minimum Drinking Age Act, which threatened withholding highway funds to blackmail the states into raising their drinking ages from 18 to 21?
I agree Jefferson was much more in favor of state systems (George Washington, to the contrary, was in favor of a national education system). But that's still not really libertarian, which would argue we shouldn't have universal free public education at all. On the state level, Jefferson pushed tirelessly for an expansive system of free public schools in Virginia (mostly unsuccessfully), and was heavily involved in setting up the state-run University of Virginia.
And I'm not sure quite how solid Jefferson's opposition to federal involvement was: he actually proposed a constitutional amendment in 1806 that would place education "among the articles of public care" so that the federal government could "come in aid of the public education".
Up until the pilot has to eject, anyway...
There seems to be some disagreement on the subject...
It depends on how you value various things. Detecting a lot of people who would not have died anyway in Phase 1 will raise your overall survival rates (those people will still live), but may make a whole lot of people's lives shittier. Is removing an X% change you'd have died worth a Y% chance that you unnecessarily made your life shitty for years? Depends on X, Y, and your preferences.