Guan’s blog

It’s even better than it appears:

1. Eating 200 bananas = chest x-ray. xkcd.
2. What the NYT paywall looks like. (Also: there are Canadian proxies?) (via Tyler Cowen). Nieman Journalism Lab.
3. The CIA’s invention of Facebook has saved the government millions of dollars. Onion News Network.
4. Smart way of mounting an iPhone on a bike. rapman.
5. “Italy is made. All is safe.” Wikipedia.
6. For industrial pencil sharpening. Gizmodo.
7. Reactionary academic journals and their MSM compatriots, Danish edition. Politiken.
8. A source of cheap 433 MHz RF Link transmitters and receivers. Seeed Studio.
9. How to dump a Chinese girlfriend (via Christine Tan). Shanghai Shiok!

Kevin Drum:

AT&T’s proposed acquisition of T-Mobile would make AT&T about a third bigger than it is now. Here’s my prediction about how this will play out if the deal goes through: within five years, the compensation of AT&T’s CEO will grow to be about a third bigger than it is now.

There's no especially compelling reason this should happen, but it will regardless.

There is a compelling reason why CEO compensation should increase with firm size: a CEO can add more value in a larger firm. This is a key element of Gabaix and Landier (2008), but it also appears in classical microeconomic models where wages are equal to the value of the marginal product of labor, which in turn (with a Cobb–Douglas production function, for example) increases in the amount of capital deployed. That’s an assumption of course, but a realistic and standard one.

Of course firms may still be too large and CEOs may be overpaid, and all this may be welfare decreasing. And the compensation of AT&T’s CEO may well grow to be more than a third bigger than it is now.

Miscellaneous tweetage from the last few days:

1. Signs of the paywallpocalypse: “The article is here (beware Canadians, not worth the click!)”.
2. Felix Salmon bait: “She was just informed by the IRS that her application would have to wait because they’ve received 4,500 applications for new nonprofits to respond to the disaster in Japan.” 4,500 new charities for Japan, hat tip Chris Blattman.
3. These guys wrote a 300/1200/9600 baud and DTMF softmodem on an AVR 8-bit microcontroller: Byonics.
4. Apple: time to move on from Commodore 64 software. ForkBombr.
5. Coasting in neutral does not save gas. Popular Mechanics, hat tip Dave Winer.
6. What the Luddites really fought against. Smithsonian.
7. So, we should eat more horse and rare pork. Alex Tabarrok.

I use Shaun Inman’s excellent Mint for analytics on this blog. The default instructions are to link to Mint’s JavaScript using a script element inside the head element. Unfortunately this means that page loading will block while Mint loads, which can add several hundred milliseconds to load time.

Fortunately Google provides some code to load Google Analytics asynchronously and this code can be easily adapted to Mint. Just put this at the bottom of your page and remove the script element for Mint from the top:

<script type="text/javascript">
(function () {
    var ma = document.createElement('script');
    ma.type = 'text/javascript';
    ma.src = '/mint/?js';
    ma.async = true;
    var s = document.getElementsByTagName('script')[0];
    s.parentNode.insertBefore(ma, s);
})();
</script>

I asked Sean, my resident expert on all things text, to comment on Anne Hathaway driving Berkshire Hathaway’s stock:

If only we knew exactly what these guys were doing. We’re pretty sure they’re using things like naive bayes classifiers to both 1) classify articles as being relevant to a particular stock and 2) predict return events from the text of news, and this is exactly the kind false positive you’d expect to see in both cases. If there isn’t already, there’s going to be a big set of papers on meta-anomalies, anomalies created from automated trading systems trying to take advantage of extant anomalies.

Meta-anomalies. I like that.

Update: Sean adds more detail.