Wednesday, July 22, 2009


Eric Wiener, on Rick Steve's show, as blogged by GetRichSlowly

There have been studies that show that people are materialistic — irrespective of how much money they actually have — people who are materialistic tend to be less happy than people who are not.

Close relationships are a better predictor of happiness than monetary wealth. “Happiness is other people,” Weiner says. “Our happiness is determined in large part by our quality and quantity of relationships with others.

Let’s talk about Denmark, for instance, because Denmark ranks consistently in the top three for happiest countries in the world. The Danes have low expectations. In survey after survey, they’re asked about expectations, and they have relatively low expectations. We Americans have very very high expectations. And I think that partly explains the discrepancy.

I think if you have low or moderate expectations, you’re less likely to be disappointed. You’re more likely to be satisfied or content. You’re more likely to be happy.

Sunday, July 19, 2009

Craftsmanship, not engineering

Post at Coding Horror, "Software Engineering: Dead?"

What DeMarco seems to be saying -- and, at least, what I am definitely saying -- is that control is ultimately illusory on software development projects. If you want to move your project forward, the only reliable way to do that is to cultivate a deep sense of software craftsmanship and professionalism around it.

The guys and gals who show up every day eager to hone their craft, who are passionate about building stuff that matters to them, and perhaps in some small way, to the rest of the world -- those are the people and projects that will ultimately succeed.

Everything else is just noise.

Friday, July 17, 2009

Statistics Resources

Cancer Genomics Tools from Washington University (WUSTL) includes a list of R functions that can be used for various statistical tests.

UCLA's Academic Technology Services department has a page with R links and information. In particular, they have a list of analyses and sample code.

Wednesday, July 15, 2009

Dunbar's Number and Security

Post by Bruce Schneier on his blog, "Security, Group Size, and the Human Brain"

In a 1992 article, Dunbar used the correlation observed for non-human primates to predict a social group size for humans. Using a regression equation on data for 38 primate genera, Dunbar predicted a human "mean group size" of 148 (casually rounded to 150), a result he considered exploratory due to the large error measure (a 95% confidence interval of 100 to 230).

Several layers of natural human group size
3-5: Clique - people you would turn to in times of severe emotional distress
12-20: Sympathy group - people with whom you have special ties
30-50: Typical size of hunter-gatherer overnight camps
150: Approximate maximum number of co-workers
500: Megaband
1500: Tribe

Note: All of these numbers have very large confidence intervals.

These numbers (and particularly the ~150 number) are important because of their effects on organizational behavior.

Coherence can become a real problem once organizations get above about 150 in size. So as group sizes grow across these boundaries, they have more externally imposed infrastructure -- and more formalized security systems.

Small companies can get by without the internal forms, memos, and procedures that large companies require; when does what tend to appear?

Tuesday, July 14, 2009

Machine Learning Resources

Stanford University CS229 (Machine Learning) Materials
  • Andrew Ng's presentation - ML Advice
  • Problem: Overfitting (high variance)
    • Diagnostic: Training error is much lower than test error
    • Solution: Larger dataset, fewer features
  • Problem: Too few features (high bias)
    • Diagnostic: Training error is high
    • Solution: Larger set of features, different features

Wednesday, July 8, 2009

Great analogies for wireless security

From a article
  • WEP is like a home bathroom lock, the one you can open just using a bent paperclip. Everyone knows how to unlock it, but when it's locked everyone who walks by understands they should stay out.
  • WPA is like a standard door lock; it's a lot more secure, but it is still possible to get by for someone with the right tools, knowledge, and circumstances.
  • WPA2 is like a bank safe. It may be possible to defeat, depending on how it's been set up, but it's not realistically possible for anybody to actually do so... yet.
  • Not broadcasting your SSID is like taking the numbers off of your house - The house is still there and everyone can see it, it's just a bit harder to find for people that don't know what they are looking for already.
  • Filtering by MAC address is like having a guard at the door that checks everyone's name against a list to see if they can enter. The only problem is, he doesn't ask for ID or remember what people look like, so anybody can and can listen in to see what names are allowed and then claim to be anybody else.