Mes Souhaites

Apparently, some hackers copied a bunch of documents from some climate change researcher's computers. I am not going to speak to that particle can of worms, doubly so since I have currently only read about it from some of the more... uh... colorful bloggers I read(The Austrian Econ/Libertarian - Beware The Ice Age Cometh: Hackers Prove Global Warming Is A Scam and this other blog - Hackers Release Private Global Warming Emails(I don't even know how to describe that one)).

And to top it off, the level of distrust all this will bring up(ignoring Fox News, I'm talking about the rest of the press, if they even notice...) reminded me of this subversive reactionary blogger I read who is discussing an "Antiversity" and how it could create a coup against democracy(uh, the guy is seriously not a fan of it): The Dire Problem and the Virtual Option, and more here and here).

Anyway, all this is an excuse to ramble about some of my "crazier" leanings. A lot of people have compared good science to open source. I actually think that's a good comparison, and this whole incident will bring up how good open source differs from bad, and how that applies to science. The best open source is done in the open: public mailing lists, blogs, public version control repositories with the complete history of the project.

To quote one of my favorite movies, its run under the rule: "No More Secrets".

Of course, a lot of projects and science are not run like this. Some of it is just human nature, after all, people fear looking like idiots if they show their works-in-progress to the world, others worry about getting scooped on things sitting right under the noses, and finally: Greed.

Untangling the greed aspect is hard, because there are so many tentacles to it: Researchers in fields with lucrative industry ties want to get rich from their research. Universities wanting to claim faculty work and even more questionable, student work(why is this more questionable? Plenty of students PAY the school to learn, at least the faculty have a salary and a formal contract regarding this stuff...).

Even worse, and this relates to my personal pacifism and complete hatred of the Military-Industrial complex: The way the government corrupts research and morality with its grant giving process. When I was still on an academic trajectory, I heard plenty of people point out that I was retarded for being completely, emphatically opposed to taking military money. And this included pacifists. The problem, of course, is that this sort of casual attitude promotes a laxness in standards and judgment, and undermines the connection between personal morality and your work.

Now, scientists should be objective, so we could argue about my desire to connect morality and ethics to the process of science. But scientists are also people, and ultimately, the scientific process is a commitment to a set of ethics and morality(if scientists should not be concerned with morality, let me invoke Godwin's law and ask about all those interesting experiments on human test subjects the Nazi's performed?)

Of course, another affront I have is that public money funds a good amount of research in this country. Now, I have suggested that good science, like good open source is performed out in the open, furthermore, computers are quite amazing at storing large amounts of data easily, so why, in this day and age, would we be concerned about not having, capturing, storing, and publicly archiving every "artifact" produced by a research team? Furthermore, why do my tax dollars fund public research which is quite frequently handed off to private hands for development, without ever having a full disclosure of the record of the research? There is even a constitutional question here, since the US federal government is not permitted to exercise copyright on anything it produces. Of course, contractors WORKING for the government can have copyrights? Legally it all makes sense, logically, however, it does not(at least that's my claim).

Now, some might argue science would look radically different if the sort of stuff I mentioned here was even partially implemented. Yep, that's my point. Good science is done in the open. Intellectual property is frequently about keeping everyone else in the dark. That's what we would call a conflict of interest. And regarding all those sweet deals between government funded research and private hands? I don't really see how its that different that other forms of corruption. I may be a serious capitalist, but I strongly believe in the value of public goods and science is one of the ultimate public goods.

Today is international toliet day, and thus, my intuition that today was going to be a crappy is confirmed.

Goldman: Flu Fear Spurs Donation!, it seems the vampire squid has required its employees to no longer use coins(also, to let their drivers always open their car doors as well, which is a good point, I will inform my driver of this idea later...).

The Origin of Quake3's Fast InvSqrt().(like I said, via Hacker News).

And funnily enough(probably why it popped up on HN), today's xkcd mentions this bit of code(hint, mouseover...)

And XKCD gets it sooooo right. Just see my previous post for an example of locked up code. Of course, I can only hope to write something that awesome one day.

(Of course, I think I might give that bit of code a whirl in the JVM just to see how it compares... a certain N-body simulator does a lot of invert square roots...).

Oh yeah, another bit of arcane knowledge. FMUL is typically 10x faster than FDIV. So its faster in most cases, if you are reusing the denominator of a calculation, to cache the 1.0 / denom value and multiply it repeatedly. Now, what will really blow your mind is figuring how Carmack timed the parallelism in the FPU to stage FDIV results to come out just in time for their use in the next 8 pixel span of a bilinear texture mapping routine, so the next divide could be scheduled while the span was drawn needing the last one. Effectively amortizing out much of the cost of the FDIV.

Smart Grid Could Pose Threat To Privacy. Yep, I've been to the future. It does pose a giant threat. Big one. I'm talking post-nuclear-holocaust-giant-mutant-cockroach big. Crooks can know when you're on vacation(And write a query to look for it), and we'll all know about your hydroponic marijuana farm.

Its all true. I didn't even read the article, the comments, or anything. Cower in fear though. Also, since most power utilities are public, I think we can do Freedom of Information Act lawsuits to get your minute by minute power utilization too.

Anyway, the only good news is that when I wrote this scary software, the company I wrote it for seemed to do a horrible job of selling it, so you have a few more years of privacy left. Enjoy it.

(Yes, just now looked at the comments, I nailed the subjects of the first two comments, pot and thieving. I guess we know what everyone at /. does...).

Of course, data mining at large scale has horrible flaws, and the world has a lot of incompetent employed programmers. Concerns are vastly overrated. First, it will take many years to replace all the meters, one by one(and replace is the keyword), and the software from most of the meter reading companies is complete crap, and really, what 99% of people will actually be upset about is the sky-high cost of on-peak power usage.

You see, the power company doesn't care about your hydroponic pot. No, seriously. What they care about is everyone simultaneously deciding to max out their AC, while taking steaming hot showers, doing laundry, and cooking at prime time in the evening. Fun fact, during peak power usage in the summer, TVA loses money selling us power.

So really, you'll just start doing your laundry at 3am to take advantage of dirt cheap power.

As if my live ultra spam news spiffy name-of-the-week feed on Facebook isn't spammed enough with Farmville, now an MMORPG blogger compares Farmville's persistence to a typical MMO.

I suppose that's my punishment for being up at this hour.

Police Used Taser on 10-Year-Old Girl in Arkansas

(And if you think a "very, very brief" tasing is okay, perhaps you should get a very, very brief punch to the face sometime too...)

The Mythology of the Future Job Market.

Normally, I would get myself worked up into a rant over this sort of article. And of course, why are articles like this getting posted? Because unemployment in the US is getting crazy high. In fact, we're not even sure who to believe, U6 unemployment(i.e. not the one reported when they say "Unemployment" in the news media) has been in the teens for a while, and the employment to population ratio is not looking so good either.

Anyway, I get annoyed because I'm not a techno-utopian(or one of the cousins/sub-genre's: singularitarians). You have more forces at work here than these articles. First off, I have met plenty of people who hate automation. People in management positions as well. Ignoring the aspects that are well-grounded(just for the moment), some of it is because there is a class of "capitalists/managers" who hate the idea of paying an engineer or programmer to automate something. These people will likely always be with us, and they will only automate against their will. And if you want to claim market forces will force them to automate, consider how "market forces" have pushed all the bankrupt corrupt plutocrats on Wall Street into the poor house.

As to the issue of automation being "efficient", well, first off, I do not think "Strong AI" is possible, and if it is, it is in the far-future(we will all be long dead before it happens). I won't go into why, and I won't be "authoritarian" and say trust me. I believe this, and have strong technical reasons why I believe it. Anyway, automation has to be maintained, and has to be changed when the underlying processes change, and it has to be tuned, observed, and managed. All these activities are far more complex than arm-chair pundits and economists assume they are. First off, some "business processes" are so non-routine, even in large businesses, that any attempts to automate will cost excessively more to implement than to just do it by hand.

Of course, financial engineers are helping with that last one. By making money cheap(0% interest rates, or the "negative interest rates"(for bankers only) of quantitative easing) they can make it cheaper to automate than pay. Of course, they will destroy all faith in the currency and economy with that, so, hopefully the Fed and friends will decide it was totally worth it.

But the problem still remains, one-off, low-repetition tasks are not worthwhile to automate at nearly any cost.

And that, of course, brings up the point that so many of these articles make: there are very few truly well trained polymath virtuoso's. And of course, it is these Randian Ubermensch who get the top of the pay scale, from the article:

Jobs that rely heavily on creativity, talent or unique personality traits (think authors, actors, musicians, commission sales people) very often have a power law income distribution. In other words, a few people do phenomenally well, while nearly everyone else struggles to survive. Even if vast numbers of workers could successfully migrate into these more creative areas (and I doubt that), it would probably do very little to slow down our drive toward ever-increasing income inequality.

First off, winner-take-all is a shitty model for any economy to take, and will slowly-but-surely drive even good people out of fields. But ignoring that "theoretical" issue, something I will observe, perhaps anecdotally, is that there is a shortage of people truly gifted at this stuff. And in the field of software, hiring a shitty programmer is quite common, and, well, results in a bad outcome and automation that frequently fails to successfully or completely automate.

So, all these articles are ignoring the elephant in the room. The economy sucks, its engineered, its not a free market, too big to fail is income redistribution toward the rich, and good old fashioned protectionism would help shut out some our "trading allies" mercantilist BS, and maybe actually help the "common person". Of course, that would require the upper echelons of our aristocracy to give a shit, and they frequently don't. So its not automation that is killing jobs, its not even government regulation, its just good old fashioned corruption and incompetence.

A Case in Antiquities for 'Finders Keepers'.

Of course, I like the idea of people having access to their cultural heritage, but I can also very clearly see the case that museums don't like shipping valuable artifacts back to their home countries which might also be war zones at the time. And then again, you have cultural imperialism and "white man's burden" issues going on as well. And frankly, while we can all enjoy Indiana "It belongs in a museum" Jones, it doesn't take much reading of history to know that many "archaeologists" of yesteryear had more in common with Pirates than Academics. Anyway, interesting to think about.

ditaa looks handy(via this post).

Why you would be wanting to convert diagrams to ascii art, I don't know...

Okay, that's unfair... I would actually probably use this myself... because I tend to keep all my notes and stuff as raw text files rather than use a word processor(not even HTML... just raw test. SciTE launches so fast you barely notice it, and no fiddling with font or formatting as Word processors are apt to encourage...).

So we were going to go to the Knoxville Zoo Free day today.

The sun was shining, the weather was warm, and the economy bad.

The traffic to the zoo was backed out onto I-40. In fact, backed out over a mile from the exit itself. The cars in this line were not moving. As we drove past the exit(due to the sudden realization that those cars were in line for it...) we saw cars parked on the hills and dales around the zoo... parked anywhere and everywhere.

Maybe admission is not so expensive?

So we did not, in fact, go to the Knoxville Zoo today.

Random Fractal Hextiles Backgrounds.

Hit reload for more pretty pictures.

Okay, so this isn't the greatest example(Winamp Milkdrop visualizer is pretty close to the top...), but seriously, why isn't there a business model for visual crack cocaine? I mean, its right there in the name, I mean, real crack cocaine has a thriving business model...

So, Context Free languages are exactly equivalent to languages that can be accepted by a Push Down Automaton.

In plain (programmer?) English, you need a stack to represent nesting, and nesting is exactly the "difference" between a Regular language and a Context-Free language.

Since Regular expressions are a language for writing down any regular language. It kinda follows that if a language is Context Free you can't parse it with regular expressions. Since Context-Free languages are not, by definition regular. If your language has nesting, also, it follows that its not regular.

In other words, you can't parse X/HTML/ML with regular expressions.

Which explains why someone asked for a RegEx match open tags except XHTML self-contained tags on Stackoverflow.

Of course, there are plenty of programmer's in this world who will laugh at "stupid theory people" like me. So they will, of course, try to use regex's to parse languages with nesting. Some people can be educated, but the rest will just ride that fail whale hard.

And yes, I wrote a whole master's thesis on parsing, thank-you-very-much.

Defense Spending and Job Losses

East TN man found asleep in ditch with rifle, moonshine.

Nice.

Do Businesses Hate Their Workers? (Income Disparity Myths Edition)

I suppose it says something about my cynicism to call this a "fun" article...

Literal Real Estate Crash.
Not like being a professional engineer(software to civil...) matters...

The local market has a ton of wonderful pumpkins that we're trying out recipes with. This week was the warty "Galeuse d'Eysines," which I think will get turned into something sweet (though the pumpkin fries weren't half bad). Luckily, these didn't get patented.

What Good Is a College Education?
I enjoyed the link. Thought you guys might too. Unfortunately, there is a dangerous negative feedback loop for society when the vast bulk of people think: "college = good job". It can, often does, and is something that will probably always be helpful for a good job, it has become a source of just as much angst: People with degree's who can't find jobs, the number of people who just serve to clog the system up with credential inflation(those skated/cheated through college because, after all, "it's just bull shit to get a job", and those are the very people who ruin classes for people who are not there "just to get a job").