Month: March 2010

My Favorite Japanese Pickle: Narazuke

Seth Roberts4 Comments

I think the Japanese have the most sophisticated fermented food culture in the world. The French have cheese and grape wine; the Japanese have miso, natto, rice wine (sake), and a wide range of pickles. It’s no coincidence, I believe, that the Japanese and French have two of the lowest rates of heart disease in the world, in spite of high smoking rates. Perhaps fermented food gives you a taste for smoking, which provides complex flavor.

My favorite Japanese pickle is called Narazuke. It is melon or vegetables such as cucumber and eggplant pickled in sake lees (the rice left over from making sake) for 1 to 3 years. Two days ago I had a pickle from Guss’ Pickles (a “New York barrel-cured sour pickle”) and a piece of narazuke. After eating the narazuke, the Guss pickle, no disrespect, tasted like it was made by a 10-year-old. The complexity of the narazuke is so much greater. Which is hardly surprising because it is aged so much longer. A Guss pickle might be aged two weeks.

Andrew Gelman’s Top Statistical Tip

Seth Roberts4 Comments

Andrew Gelman writes:

If I had to come up with one statistical tip that would be most useful to you–that is, good advice that’s easy to apply and which you might not already know–it would be to use transformations. Log, square-root, etc.–yes, all that, but more! I’m talking about transforming a continuous variable into several discrete variables (to model nonlinear patterns such as voting by age) and combining several discrete variables to make something [more] continuous (those “total scores” that we all love). And not doing dumb transformations such as the use of a threshold to break up a perfectly useful continuous variable into something binary. I don’t care if the threshold is “clinically relevant” or whatever–just don’t do it. If you gotta discretize, for Christ’s sake break the variable into 3 categories.

I agree (and wrote an article about it). Transforming data is so important that intro stats texts should have a whole chapter on it — but instead barely mention it. A good discussion of transformation would also include use of principal components to boil down many variables into a much smaller number. (You should do this twice — once with your independent variables, once with your dependent variables.) Many researchers measure many things (e.g., a questionnaire with 50 questions, a blood test that measures 10 components) and then foolishly correlate all independent variables with all dependent variables. They end up testing dozens of likely-to-be-zero correlations for significance. Thereby effectively throwing all their data away — when you do dozens of such tests, none can be trusted.

My explanation why this isn’t taught differs from Andrew’s. I think it’s pure Veblen: professors dislike appearing useful and like showing off. Statistics professors, like engineering professors, do less useful research than you might expect, so they are less aware than you might expect of how useful transformations are. And because most transformations don’t involve esoteric math, writing about them doesn’t allow you to show off.

In my experience, not transforming your data is at least as bad as throwing half of it away, in the sense that your tests will be that much less sensitive.

How Long To Build a Subway?

Seth Roberts4 Comments

From New York:

1972: Workers use cut-and-cover to break ground on the Second Avenue subway line; only a mile of tunnel is completed before the seventies financial crisis halts construction. . . .
2007: Ground breaks once again on the Second Avenue subway, to be called the T line.

Slow, yes, but not  off-the-charts slow:

Between 1965, when construction of the Beijing subway began, and 2001, workers laid only 42 km of track.

Faster is possible:

By next year, Beijing aims to have another 100 km of track up and running.

As I searched for Beijing subway info, I came across this surprising blog on the Beijing City Government English website. It reminds me of something that happened to me in Alaska. I went to visit a glacier. Near the glacier was a visitor’s building, which had a small room with a slide show. The taped narration told how the glacier shrank during the summer and grew during the winter and described the animals that lived nearby. It was all very bland but you could tell the narrator really cared about the glacier. I was struck how rare that was: To see that someone really cared about something other than themselves and their family. I suppose this is why I was impressed how much Penn State students love Penn State. This blog gave me the same feeling. The writer likes (or rather liked) living in Beijing and, much more impressively, manages to convey that. I nominate it for best blog on a government website. Unfortunately it has stopped. It’s so much easier to learn when the person you are learning from really cares about the material. There’s a lot I can learn from that blog.

A nice video about building Beijing’s subway.

A Chinese Farmer Fights Back

Seth Roberts1 Comment

This is from China Daily:

Every day before sunrise, Zhang Zhengxiang leaves home to walk along Dianchi Lake, one of the major attractions in Yunnan province.

The 62-year-old retired farmer carries a camera, tripod and telescope to record the pollution encroaching on the country’s sixth-largest freshwater lake.

During weekends, Zhang collates his observations and sends letters to the local government, informing them of the growing pollution.

He has been doing this for 30 years.

Sounds good to me. Like my self-experimentation, he is (a) trying to change something he cares a lot about and knows a lot about and (b) slowly collecting data. In contrast to a great deal of American good works, such as Jeffrey Sachs’s.

In this case, unlike a lot of philanthropy, we know how the story ends:

His efforts slowly began to pay off.

In 1998, the local government shut down six mines near Dianchi because of his warnings.

In 2003, 56 large and medium-sized mines, chemical factories, and fertilizer and lime plants were closed.

Since 2008, the local government has invested about 12 billion yuan ($1.7 billion) to clean up the lake. . . .

[In 2005], Zhang was selected as one of 10 outstanding grassroots environmental activists. In 2007, he became a member of the Chinese Society for Environmental Sciences.

Last year, he was selected as one of the 20 people who have warmed Chinese hearts.

This supports what Jane Jacobs told an interviewer: “It’s a funny thing. You can only change something if you love it.”

Success with the GAPS Diet

Seth Roberts2 Comments

Darrin Thompson writes:

Thanks for the pointer awhile back to the GAPS diet. It caught my eye and my wife and I implemented it for us and our autism kids. After about a month we are experiencing marked reductions in psoriasis and allergy symptoms. Our 3 autism kids are [now] doing well with no huge barrages of expensive vitamin supplements. We’re keeping up only with Vitamin E, DHA and eventually selenium. We’re noticing improvements in their communication skills.

Why I Use Arithmetic to Measure Brain Function

Seth Roberts8 Comments

At the latest Quantified Self Meetup, I described my recent arithmetic results. Gary Wolf asked why I used arithmetic to measure brain function.

The simplest answer is that it won a competition. I tried eight or so different ways of measuring brain function and it was the best. I had high hopes for all of the tests and spent a lot of time on most of them. I hoped to find a test that was as fun as playing a game yet provided the detailed and controlled info of a typical cognitive psychology task. I didn’t quite get there.

The other tasks had several problems:

1. Long learning curve. My belief that our environments might be having big and unnoticed effects on how well our brains work began when I noticed that my balance was better the morning after taking 6 flaxseed oil capsules. A small amount of flaxseed oil substantially improved how well my brain worked — at least on one dimension. The test I devised to measure this effect confirmed its existence but after doing it for a year my score was still improving. Obvious learning during testing makes it harder to interpret the results — you need to correct for the learning. I wanted a task where there was less room for learning. I already know arithmetic well so an arithmetic test took less practice until learning stopped.

2. Not so portable. Some of the tasks required props (e.g., a printed sheet of paper for a paper-and-pencil test).

3. Trial-to-trial interference. Wanting to cram as many trials as possible into a short time, I started with an arithmetic task with 5 blocks of 20 trials each. There were pauses between blocks. During a block, however, as soon as I finished one question (e.g., 3+4), the next one would appear. At first, this was fine. As I got faster, though, I started to make a large number of mistakes quite often. I required 85% accuracy and it started to take longer and longer to do 5 blocks with acceptable accuracy. The arithmetic task I use now has separated trials. Instead of 100 trials in 3 minutes I do about 32 trials in 3 minutes, but there has been no increase in the amount of time they take. (I found that 3 minutes was as long as I would comfortably do the test; 32 is simply the number of trials I can do in 3 minutes.)

4. Carpal Tunnel Syndrome. After doing a lot of an earlier task that involved just two fingers, my wrists started to hurt. The arithmetic task uses eight fingers (excluding the thumbs) equally. 1, 2, 3, 4, 7, 8, 9, and 0 are possible answers; 5 and 6 are never answers.

5. Subjective. One of the tasks I tried involved making a single motion on the touch pad. It wasn’t easy to be sure I’d made one motion rather than two.

6. Small number of measures. One task I tried only generated five scores in four minutes. The arithmetic task I use generates 32 scores in three minutes. Each of the 32 scores can be corrected for the difficulty of the problem (e.g., 0+0 is faster than 6*7).

7. Unpleasant outcomes. Some of the tasks were frustrating. When I didn’t do well it was irritating. For some reason that doesn’t happen with the arithmetic task.

A more game-like task would be better but for that I may need a new input device, like a joystick. R — the language I use to collect the data — is great for data analysis but poor for data collection. it is unlikely to work with joysticks anytime soon. My arithmetic task needs an R function that only works with Windows, which shows how bad the problem is. It’s a miracle that R function exists. Maybe it was written by an experimental psychologist.

Even More Astonishingly Bad Dreamhost Support

Seth RobertsLeave a comment

Dreamhost supplies a way to backup your website. You click a button, a few hours later you get an email telling you to go to a certain place to download a set of files. Those files are supposed to be the backup. That’s what I did. I got a message that said:

Ta da! Â Your ENTIRE DreamHost account has been backed up now here: . . .

Note the emphasis: ENTIRE. Now it appears those files don’t work. They aren’t a complete backup of my website!!!

In spite of this astonishing fact, someone in Dreamhost Customer Support told me ” We havent had any users report issues with the backups that we create.” Amazing. No users ever!

I want to hire someone to fix the problem. If you have the technical skills to (a) repair the SMF forums (which will require upgrading) and (b) help me transfer to a new hosting service, please contact me. You can reach me at twoutopias ..[at]…. gmail.com.

More Someone kindly offered to help me and the problem has been fixed. I still hate Dreamhost. When I complained bitterly about their bad backup, in reply they sent me an email that implied it was my fault! If you send me your stories about bad experiences with Dreamhost, I would be happy to post them.

Assorted Links

Seth Roberts1 Comment

Thanks to Carl Hattery.

Science in Action: Mysterious Mental Improvement (part 4)

Seth Roberts21 Comments

I blogged earlier how I suddenly got better at an arithmetic task. The apparent causes of the improvement were butter and standing. I’m not sure this is right; I will do more tests.

While I was trying to figure out the cause something even more extreme happened:

Notice the last two points. The previous anomaly was slightly below 600 msec. The new one is close to 550 msec. After observing it, I repeated the test 20 minutes later and got essentially the same result.

I’m blown away. I’ve been doing tests like this — simple measures of mental function — for about two years. Nothing like this happened during those two years.

My scores on this particular test averaged about 640 msec. Sometimes they’d be lower (as low as 610) but I had no idea why. The average stayed around 640. Now, within days, the average goes down to about 600 (presumably because I was eating butter regularly) and then down to almost 550. In other words, that 640 could be improved almost 20%! The improvement has nothing to do with practice; I was extremely well-practiced on this task. (And practice doesn’t produce such a sudden improvement.)

This is something we care deeply about — how well our brains work. Unless I’m a lot worse at arithmetic than everyone else, this suggests that for many people great improvement is possible. In an astonishingly small way. (I didn’t make any big changes during this time.) In a week.