2007-10-31

Oil peaked, but uranium can last us 'tens of thousands of years'

New Scientist relays an announcement by the German Energy Watch Group of a study showing that oil has peaked and that production will now decrease by 7% annually, halving by 2030. They also predicted "significant falls in gas, coal and uranium production".

This short 2004 article by James Hopf argues that the price of uranium contributes a radically lower share to nuclear power generation costs than is the case for coal or oil. According to Hopf, the cost of uranium ore could easily increase to $1000 per kilogram while contributing negligibly to the overall cost of nuclear power production. He quotes the 2004 price as $40/kg, and though it has bubbled considerably since then, the peak was $300 per kilogram in June, with the current price around $190.

The point Hopf makes is that uranium is abundant in sources such as granite, and even in seawater, so although extracting uranium from these sources may be costly, technologies such as breeder reactors make the acceptable cost threshold very high indeed - perhaps as high as $50,000 or $100,000 per kilogram in today's dollars.

Combine that with the fact that the ecological footprint of nuclear power generation is much lower than that of any of the "sustainable" technologies, and I think we have a pretty good - and green - solution to our energy needs indeed.

And not just that - with peak oil supposedly already having occurred, and with China and India rising, I don't see how there is even any choice: the power will have to be generated somehow, and after a relatively short while, the only remaining viable "somehow" will be nuclear power. Other "sustainable" methods cannot cope unless we dedicate the surface of the Moon to solar farms and somehow connect it to Earth by cable.

2007-10-29

Thoughts on development in St. Kitts, and some photos

We just love it here in St. Kitts. It's been almost 6 months since we arrived, and our experience has been decidedly optimistic. The island is brimming with development; it's hard to take a photo anywhere around Frigate Bay without capturing a site under construction. Much of this appears to be in anticipation of Christophe Harbour, a major development which is about to start on the south tip of the island. Initially, according to the recent announcement, there will be a mega-yacht harbour; two luxury hotels; and one more golf course. (Marriott already has one in the area where we live.)

This is great news for St. Kitts. On the one hand, development on this scale is a blessing; on the other, it is also urgently required. The government of the past 12 years has amassed a national debt exceeding US$1 billion, and the 2005 GDP of US$453 million is insufficient to allow a debt of this size to be gracefully repaid. In order to repay the spending of the past decade, the country would either have to tighten its belt considerably, which would have serious repercussions economically and politically; or it has to increase its GDP considerably, so as to allow the debt to be repaid without dangerously increasing taxes.

Hopefully, with the development of Christophe Harbour - along with investment in the rest of the island that dovetails it - the latter will occur; and there is the prospect that, with all the infrastructure and all the new people, the island is going to become a pleasant place to stay, indeed.

Then, all we have to worry about are natural disasters - as opposed to economic ones that are man-made. :)

Here are some pictures we took today. As one might conclude from the presence of the cruise ship, the tourist season is beginning. The other two photos are from the same hill, in different directions.

(Links to previously posted photos: May; October)





2007-10-27

The Russians

There is much to be learned from people who occupied positions in the intelligence structures of communist countries, and chose to defect to the West. Unlike any other personal profile, these people combine (1) access to extremely privileged information - about facts, tactics, and strategies; and (2) a moral spine which eventually drove them to defect, as well as to speak publically and honestly about their experiences.

One of these people is Ion Mihai Pacepa, who since his defection in 1978 has published several books and in recent years written several fascinating articles, e.g. in 2006 The Kremlin's Killing Ways.

Another of these people is Yuri Bezmenov, apparently a Soviet intelligence agent who worked in India before defecting to the West in the late 1960s. An interview with him was released in 1984 by G. Edward Griffin, and Google Video features some very tasty bits of this interview. Another, shorter video on YouTube contains more background information but the editing is more patronizing and includes less of the tasty bits, too.

Here are some quotations from the interview:
It's a great brainwashing process which goes very slow and is divided in four basic stages. The first one being demoralization; it takes 15-20 years to demoralize a nation. Why this many years? Because this is the minimum number of years which requires to educate one generation of students in the country of your enemy, exposed to the ideology of your enemy. In other words, Marxism-Leninism ideology is being pumped into the soft heads of at least three generations of American students. [...] The next stage is destabilization. [...] This time - and it takes only from 2-5 years to destabilize a nation - what matters is essentials: economy; foreign relations; defense systems. And you can see quite clearly that in some areas, in such sensitive areas as defense and economy, the influence of Marxist-Leninist ideas in United States is absolutely fantastic. I could never believe it 14 years ago when I landed in this part of the world that the process will go that fast. The next stage of course is crisis - it may take only up to 6 weeks to bring country to the verge of crisis, you can see it in Central America now - and after crisis, with a violent change of power, structure and economy, you have, so called, the period of "normalization"; it may last indefinitely. Normalization is a cynical expression, borrowed from Soviet propaganda - when the Soviet tanks moved into Czechoslovakia in 1968, Comrad Brezhnyev said: "Now the situation in brotherly Czechoslovakia is normalized." [...] Your leftists in the United States, all these professors and all these beautiful civil rights defenders - they are instrumental in the process of subversion, only to destabilize a nation. When their job is completed, they're not needed any more. They know too much; some of them, when they get disillusioned, when they see the Marxist-Leninists come to power, obviously they get offended. They think that they will come to power - that will never happen of course, they will be lined up against the wall and shot. But they may turn into the most bitter enemies of Marxist-Leninists when they come to power. And that's what happened in Nicaragua - you remember most of these former Marxist-Leninists were either put to prison, or one of them split and now he's working against Sandinistas. [... Recounts more examples in Grenada, Afghanistan and Bangladesh ...] It's the same pattern everywhere.
Highly recommended. There's lots more.

Incidentally, it appears that Pacepa thinks it 'perfectly obvious' that WMD existed in Iraq, but were hidden with Russian help. He notes that a similar plan was prepared by the Soviet Union for Libya, and that such a plan existed and was implemented in Iraq.

If this is true, it actually helps my perception of the world make more sense, and it also changes the way I view George W. Bush dramatically. I always wondered why Tony Blair, seemingly an intelligent and clear-minded individual, one not as obviously prone to manipulation as American media led us to believe about Bush - why would he follow Bush into this bad idea of the Iraq war if the weapons of mass destruction did not actually exist? But if the Russians hid those WMD, of which they are obviously entirely capable, then it makes perfect sense; the WMD existed, Bush and Blair invaded, but in the mean time the WMD disappeared and they were made to look like fools.

Given what Pacepa and Bezmenov tell us about Soviet strategy - and given that KGB is now obviously in charge in Moscow - disinformation lives. Perhaps it's as alive as ever.

Prize for Achievement in African Leadership

Good work, Mr. Chissano, and great job, Mr. Ibrahim. I hope this works.

See also: You get what you pay for, or Thoughts in favor of much better compensation for elected officials.

2007-10-26

Karl Dahlke and food additives

Today I stumbled upon Karl Dahlke's personal website. I found it while searching for information on how blind people interact with computers, which led me to his article on the value of the command line interface. I was impressed by his methodical approach in interacting with computers as a blind person - he has programmed his own directory listing utility as well as an editor-cum-browser to navigate the net. I went on to read other articles on his website, and I found his articles about his family's experience with foods and food additives to be of particular interest. It's fascinating stuff, the more so because he isn't indulging in panic fits and throwing around irrational arguments mostly centered around "going back to nature" (whatever that means) like so many people who argue against anything "artificial" (so bread grows on trees?); but he is instead methodically documenting his own, his wife's and his children's experience with certain foods. For each instance, he documents how he came to realize that there's a connection between a certain problem and a certain food or dietary habit; and how he worked to identify the culprit, which sometimes was straightforward, and sometimes required several years of trial and error before the real cause was found.

Not everyone has sensitivities to wheat, or corn, or dairy, or food additives and preservatives, or salicylates; but then again, many people may have problems that may be related to foods and they don't realize it. And when they do realize it, they go and have overreactions, suddenly seeing problems everywhere.

Many of us are likely to be affected by a food sensitivity problem at some point in our lives. There are people in my wife's family who needed strict diets to get better - oh, and BTW: it's not like any doctor pointed out the need for them. In one case, the doctors actually caused the sensitivity with sledgehammer antibiotic treatments, but then they had no idea how to treat the sudden onset sensitivity to what seemed like all food. The patient had to find help elsewhere to come up with a diet that, over the course of several months, helped him recover and eventually resume eating normally.

I find the totally artificial Kraft Fat-Free Cheese Singles to be a godsend - a single slice has 30 kcal and 5 grams of protein. It's also full of all sorts of preservatives, artificial colorings and additives, like much of the other stuff that's in our fridge. I believe I have been blessed not having to avoid all these processed foods - it's easier that way. But for those times when there's something in your food supply that's actually the reason why you're having a reaction, I'm glad that people like Karl Dahlke have been beating the path that most doctors refuse to - identifying the substance or dietary habit that is hurting a particular person, and eliminating it.

The double blind experiments employing many test subjects, the kind that I and science love, don't seem to work so well when a specific problem only seems to affect one person, especially if only at a certain time.

2007-10-25

The fallacy of stagnating middle-class income

The Economist's Free Exchange blog refutes the common proposition claiming that income disparity has grown while middle-class income has not increased over decades. They quote Terry J Fitzgerald, senior economist at the Minneapolis Fed:
Rather than falling by 4 percent over the past 30 years, average hourly earnings have actually risen by 16 percent. Growth in the median hourly wage went from 12 percent to a more respectable 28 percent.
Large gains at the top end of the wage distribution might seem to be accompanied by flat wages at the bottom, but that is not the case. Wage gains at the lower end of the distribution held up fairly well. Wage growth rates at the 10th and 20th percentiles were only slightly below the median growth rates, increasing by 17 percent and 18 percent, respectively. While these data confirm that wage inequality increased since 1975, they also confirm that a broad swath of middle America experienced notable hourly wage gains.
While Russ Roberts says:
In 1970, according to the American Housing Survey (from HUD and the Department of Commerce ,then called the Annual Housing Survey, Table A-1, p. 32), 36% of the 67 million households in America had air conditioning, 11% had central air. This is the earliest data available from this survey.

In 2005, the most recent data from the same survey, (Table 2-4, p. 66) 82% of the 15 million households with income below the poverty line had air conditioning, 52% had central air.
I've long been saying that Americans don't know what "poor" is. The average poor person in the United States - the kind that's on social welfare - has a standard of living comparable to a middle-class person living in Slovenia today.

The difference is, the people in Slovenia feel smug about themselves. Most of them have never been to the States, yet a majority appears to believe the Delo ("Work", "Labor") newspaper as it tells them how "bad" income disparity in the United States is, and how "bad" the poor Americans have it.

This reminds me of an anecdote from Ion Mihai Pacepa's Red Horizons. Romanian dictator Nicolae Ceauşescu visited the U.K., and as part of his visit, he was taken to a department store, perhaps Harrods. He was impressed by all the goods available - but he found it impossible to believe that what he saw was the normal situation in a department store; it was impossible to find a store with shelves so full in Romania. He found it insulting that the authorities must have stuffed Harrods with goods they must have obtained on purpose from all over! How dare they think he would be misled like that!

Not to be outdone, when foreign dignitaries later visited Romania, he had all the paltry shops in Romania stripped of all the goods that could be found, and he had them all put into one store. Then a tour of that store was organized for the foreign dignitaries. See, in communist Romania, we have goods, too!

2007-10-24

Cool stuff: skrbl

I was thinking it would be cool to have a simple, easy to use website that you could use as a whiteboard - some place you could go with your browser and type stuff for others to see, allow others to edit it, and immediately see their edits. It seemed like this would be a useful tool for teams collaborating over distances, or even conducting an online programming job interview - talk to the person you're interviewing via Skype, telephone, or chat, and have them write an algorithm on the virtual whiteboard so that you can see them as they're progressing. Beats having to shell out for hotels and airplane tickets, especially if it's for someone who doesn't really know how to code. :)

A quick web search reveals that such tools already exist, and I tried a few of them. The best one for my needs turned out to be skrbl. It's straightforward - creating a new whiteboard session takes just one click and requires no registration. Anyone can join simply by sharing with them the session URL or session ID. It works in IE6 and IE7 (as opposed to not working in them), it doesn't require Flash, it isn't heavy on the CPU, and it's responsive. Great job!

Other sites I wasn't as satisfied with included Vyew, which is almost as easy to use as skrbl, but is less responsive, makes it less easy to manage and edit text, tends to lose characters on input, uses Flash, complains about IE7 on Vista, and badgers the user to install Firefox instead. Meh.

Also interesting is GE's Imagination Cubed, which is simple and easy to use enough but requires Flash, is much more drawing than text-oriented, and starts to burn an awful lot of CPU once you've drawn a big enough picture.

2007-10-22

The Economist's blog on Newsweek happiness

The Economist's Free Exchange blog publishes an excellent post about the bullshit that liberal arts graduates who found jobs as newswriters try to sell you in mainstream media - specifically it addresses the "Why money does not buy happiness" thesis that's frequently being sold. The Economist writes:
As Paul Ormerod and Helen Johns note in their outstanding and completely non-confused monograph, "Happiness, Economics, and Public Policy", the trend in average self-reported happiness correlates well with almost nothing. Increasing inequality, for example, has also done nothing to the happiness trend. (Why don't we hear more about this?) They find a weak statistically significant positive correlation with happiness and higher crime. Yeah, weird. They also note that the variance in average self-reported happiness is often greater within a given year than between years. This is all suggests that the time-series data on average self-reported happiness contain very little useful information about anything of interest.
Read the whole thing. It contains lots of other useful information.

In general, I have come to detest the editorial and writing population incumbent in most written media. They all too often come across as naive and simple-minded socialist idealists who are trying to sell people their own illusions about how the world should work. It's as if they see it as their mission to use their position as newswriters to "help make the world a better place" - except that they are making it much worse by perpetuating idiocy and delusion.

2007-10-20

Technical: OctAlloc improved

I wrote previously about my attempts to engineer a concurrency-friendly memory allocation algorithm for Windows platforms that would be efficient for the most demanding multi-threaded and multi-core use.

I made available in the previous post an implementation, OctAlloc, which performs way better than the default heap allocator in Windows XP or 2003, and fares well against the Vista allocator in highly concurrent situations, but is less performant than the Vista allocator for single-threaded multi-core use.

I have since made two optimizations to OctAlloc which improve its performance beyond the Vista allocator in all respects. The optimizations are as follows:
  • The previously posted version of OctAlloc uses thread-local storage to store the most likely CPU number that the current thread is running on. The TlsGetValue() and TlsSetValue() operations this requires are however somewhat expensive, and do not pay off when there's only a single thread. The new version uses a less reliable, more heuristic, but also highly more efficient strategy by using GetCurrentThreadId(), an inexpensive call, to choose the allocation buckets for the executing thread. This improves performance by some 25%.
  • The previous version of OctAlloc always pops a memory chunk and returns it back to an atomic list after allocating a unit for it. This requires two synchronized operations for every call. The new version keeps handy an additional list of preallocated units which do not require popping and returning a memory chunk. This requires only one synchronized operation for most allocations.
My measurements for the new version follow at the bottom of this post below. According to my measurements, OctAlloc outperforms the Vista allocator in performance as well as tightness in managing memory.

Note that the performance figures are not comparable directly to the measurements I posted in the previous OctAlloc article. Previously, the test program kept the load-per-thread constant regardless of the number of test threads, causing 8- and 16-thread measurements to run very long. The new version of the test program varies the load-per-thread depending on the number of threads and the number of available CPUs. An algorithm with no contention should now complete in the same time regardless of the number of threads; if it takes longer to complete with multiple threads than it does with one, the difference is attributable directly to contention.

Download the new test binaries and the source code:

OctAlloc2.zip (51 kB)

New photos since May

It's been several months since I last posted our photos. We gathered a small collection since then, the largest part of it during our trip in August to Las Vegas. Here they are.

St. Kitts




Las Vegas & the Grand Canyon




Our apartment - building progress: June 3





September 8




October 7