A recent post at Signal vs. Noise asking about the merits of Rosetta Stone brought a number of entrepreneurs out of the woodwork to promote their own language learning tools.

Livemocha.com is a social network idea. The site provides free lessons, matches you up to chat with native speakers, and lets the community correct each other's writing and speaking samples. Available for 11 of the biggest languages.

Edufire is a marketplace for individual tutoring by live video chat. You select what you want to learn and then see a list of available tutors, with their hourly rate and amount of experience on the site.

Byki (Before You Know It) is software that appears superficially similar to Rosetta Stone, but they offer a free lite version, the full software for $50, and over 70 languages.

Universed collects resources for learning languages, organizing them by media type: bookmarks, videos, podcasts, and photos. I'm not sure precisely what the photos are supposed to do for you.

LingQ (pronounced "link") is a service with multiple price levels from free to $80/month, that also involves buying entire courses from personal tutors. It's based on the founder's own methods, which appear to involve things like marking the words you don't know in passages, getting back flash cards with those words, and then submitting writing samples to the tutor employing some of those words. Ten languages are offered, and there is supposed to be a community component as well.

Side note: It's problematic when a site has only a few public pages that seem totally disconnected from the real product, as if you are on the other side of the wall from the promised land. If you offer some level of service for free, there should be an easy way to try it out or demo it, without having to 'sign up.' Also, if you're claiming to have a community, you should offer some evidence of its actual existence, like Livemocha, which has random snapshots and user names on the front page.

Mango Languages offers Flash-based lessons in 10 languages. They have free and paid levels, and "Mango to go" for $150 which lets you download mp3s and PDFs. I'll bet they're working on an iPhone version, as the only portability the current product seems to buy you is not needing an internet connection.

Unilang is another community-based free site, similar to Livemocha, but with a wider variety of resources such as lessons, phrasebooks, videos, etc., in some cases created by users or adapted from other sources. With a quick look around I saw courses in Ainu, Catalan, Esperanto, and Mongolian, as well as more common choices.

Popling, for the ADD set, delivers pop-up vocabulary questions every so often, each of which you can choose to answer or ignore. Not a bad idea. There are seven languages plus chemistry and geography. You can pay $20/year for a few more features.

A lot of people also suggested Pimsleur audio courses, which I've never tried. Oh, and happy new year!

The Times Magazine discusses the debate over Value at Risk, or VaR, a popular across-the-board measure of risk on Wall Street, and the area in which I currently work. I haven't read "The Black Swan", but Taleb certainly comes across as kind of an ass in the article. It's perfectly plausible to say that "improbable" adverse events are a lot more frequent than most people realize. It's another thing to claim that VaR is useless, or that risk managers are being foolish.

The article mentions that Taleb has his own hedge fund, the ominously named Black Swan Protection Protocol, and has made "a killing" in each of the last three crises, achieving returns of 65% to 115%. But how much has he made in between? One assumes that he either goes short the market, or buys safe assets like treasuries, or some mix (I think there would have to be some of the former, or significant leverage, to achieve those returns). The S&P 500 is currently up about 260% since 1987, when he made his first killing.

There are a lot of perma-bull analysts, and fewer perma-bears. The historical returns of the market explain the population gap. And when there is a crisis, the perma-bears are trotted out on CNBC and the bookshelves as Cassandras, because they take the generally less popular viewpoint. But it's truly rare to find someone who was a bull at the right time and a bear at the right time, even for one of each cycle.

So Taleb doesn't seem to be improving on any current models simply by being bearish. As the article says, assuming that a disaster is always right around the corner would prevent you from making any investment. Obviously many risk managers made fatal mistakes in the current crisis, and there are plenty of places to assign blame. But it may well be that the optimal strategy is to assume that tomorrow will be more or less like today, even if you know that sometimes you will be wrong.

Update: my colleague Victor tells me that Taleb's strategy is most likely implemented using long-dated, deep out-of-the-money put options. Or in layman's terms, insurance policies that his theories indicate are cheaply priced (due to the chronic tendency to underestimate the likelihood of crises), and which pay off handsomely when crises occur. Until that happens, he would lose money little by little.

This pattern of returns probably takes a strong investing temperament, but it does seem compelling, and might catch on as a way to allocate some of one's money. Of course if it does, the supposed mispricings that he is advantage of could disappear.

This New Yorker-sized piece on Zimbabwe by Columbia professor Mahmood Mamdani in the London Review of Books aims to give a different perspective on Robert Mugabe than we normally get in the Western press. He goes in depth on the factional struggles after independence in 1980, and the circumstances leading to the takeovers of white-owned farmland. It's hard to accept anyone going easy on Mugabe at first, but it's a good exercise to question our assumptions about widely demonized figures.

Meanwhile, the country has just introduced a $50 billion note, and FP Passport worries that this and other factors are constricting the flow of information about just what's going on there.

I highly recommend watching Disneyland Dream, the home movie/documentary made by Robbins Barstow in 1956, and recently admitted to the National Film Registry. It documents his family winning a trip to the newly opened park, and is chock full of 1950s goodness, besides being just a well made and entertaining movie. The family all wear monogrammed Davy Crockett jackets hand made by Meg, the matriarch, and board a "Super Constellation" plane, looking tiny by modern standards, which has to stop in St. Louis to refuel and change crews. The narration fits perfectly, despite having been recorded forty years later, in 1995.

If that's not tempting enough, it turns out that a child handing out guides in one shot is none other than an eleven year old Steve Martin, as reported by the family's home town paper, the Hartford Courant.

One last thing: I never knew that 3M, the sponsors of the contest, stood for Minnesota Mining and Manufacturing company.

Over at Subtraction, Khoi Vinh and his merry band of commenters run down all the annoying details in Apple's dropping of Digital Rights Management from the iTunes store. To wit:

1. You must pay a fee of 30 cents per song to upgrade your existing library to freedom.


2. This despite the fact that many of the same songs will now be available DRM-free for only 69 cents.


3. You must upgrade your entire library or none of it, with the total fee coming to $100 or more for many people.


4. The fee will include any songs you may have bought and then deleted in the past!

Talk about a bitter pill. I have to say I'm glad I never bought anything from the store, but at the same time I never expected they would pull something quite like this.

I found out via Language Log that the new Fox show "Lie to Me" features Paul Ekman's Facial Action Coding System as a way of detecting lies. Ekman is serving as the show's scientific adviser and keeping a blog in which he takes short clips from the show and explains how they illustrate his system. (Evidently he had a real creative burst when he wrote a dozen entries at 3am last Wednesday.)

The blog is an effective demonstration of how Ekman's techniques are meant to be used. It also makes the show look a bit silly, because as in much expository dialogue, the characters tell each other things that they should both already know if they are experts.

Last time we checked in with Ekman, I was still bemoaning the paucity of video examples of microexpressions. The examples on the blog help his cause, and the folks at Language Log seem to respect him. But I would still like to know more about how common microexpressions are in the real world, and better tests of whether they can really be seen and used for inference.

I made a pleasant discovery yesterday while looking up the word salmagundi ("A dish composed of chopped meat, anchovies, eggs, onions with oil and condiments", and by analogy any assorted mixture). The full archives of the magazine, while accessible only to subscribers, can be searched by full content on Google. This is a great leap over the DVD-based Complete New Yorker, which only had keyword search, and the New Yorker site itself, which only allows you to search within the current issue.

Try it by adding "site:archives.newyorker.com" to any query. For example, salmagundi. You can see from the snippets that the scanning job isn't perfect, but still, pretty great.

Via WFMU's Beware of the Blog I've discovered the writings of Francis E. Dec, a fine practitioner of the art of insane rants that fill every square centimeter of a page, sent randomly to thousands of people and businesses. He was probably schizophrenic, but as the site devoted to him says, he also had a way with words.

In weaving his surreal worldview Dec likes to string descriptive words together: "laser beam knife robot arms" and "proliferated degenerating parroting puppet con artist..." (that one goes on for quite a while). And then there is his central concept, the Worldwide Mad Deadly Communist Gangster Computer God. The style viscerally conveys his panic and paranoia. It reminds me of Radiohead or Flaming Lips lyrics, and inspires me to write some of my own. The illustrations that have been made to go along with the writings at the above-linked site also had me laughing pretty hard at work.