Hah, good for you. They can move the train operators to other lines that could use more service!

For those who don't live in NYC, there's a shuttle train with three tracks that goes from Times Square (42nd and 7th) to Grand Central (42nd and Park/4th), which is about 1/2 of a mile/1 km. The shuttle doesn't connect quickly to any other train, and they run about every 3 minutes during rush hour and take 2 minutes. It would take less than 15 minutes to walk that distance. Instead, they should replace those tracks with moving walkways. Instead of taking 5 minutes (or more off-peak) and the cost of running subways, paying conductors, etc., it could take 7 minutes, have a lot higher capacity, and a lot lower cost. Win-win.

I don't understand how this could possibly work. Web posting and bandwidth is not free. The only reason FB is is because of its advertising and other tie-ins, none of which would work well if they weren't targeted. No web site could get substantial portions of the world's population on board if they had to charge money to cover development and server costs. If you can come up with a non-evil social networking business model that allows your platform to attract and support hundreds of millions of simultaneous users, dominate the single largest sector of the Internet, and the resulting costs, with 99.99% uptime, you're not just a genius, you're a god.

Incidentally, R does not really support 64-bit integers, either. http://permalink.gmane.org/gmane.comp.lang.r.devel/17281

People, at least some people, are happy to pay for news services, especially if they're non-profit news services like National Public Radio. (Which, at least prior to the recession, was growing rapidly!) If journalism as a whole can put itself into a realm where people feel like they're supporting something they participate in and believe in, it can succeed in the Internet era. I wrote about this a couple of months ago: http://www.harlan.harris.name/2009/10/online-publishing-micropayments-and-warm-fuzzy-feelings/
Excerpt:

But I think there’s a way that might work, a way that leverages human psychology. People like to feel like they’re in control, and they like to feel like they have a voice in the system. Micropayment systems that require you to pay 10 cents to read an article, based on a headline or a link, or subscription systems that take your money and give you something you can get elsewhere for free, just make you resentful. So instead, design the system so that you associate feeling good about what you have just read with giving money to the people who produced the content.

I know. This whole line of research is nearly impossible to make work. You'd need lasers to project coherent light onto the cornea (ow), since otherwise you can't get an image in focus, and a high-powered sensory-computational system to constantly shift the projected image in the opposite direction from saccadic and microsaccadic eye movements. I wish it were being developed by a public company so I could sell it short.

Reading the article, it seems as if the flu vaccine is reasonably effective in reducing morbidity (incidence of infection) among the majority of the population, who are healthy and have noncompromised immune systems. But, the evidence is unclear as to whether it reduced mortality (death) among people who are old or otherwise have weakened immune systems. Even if the vaccine does nothing at all for the elderly per se, it doesn't mean immunization of the healthy is a bad idea, for two reasons. First, as mentioned in the article, herd immunity effects can reduce the incidence of flu in the elderly, thus indirectly reducing mortality. Second, influenza sucks and reduces productivity by knocking people out of work for 3 days. In an economic sense, it is totally worth doing, even if it doesn't reduce deaths at all. The conclusion of the article should be "flu vaccinations are worthwhile, if not exactly for the reason you thought they were."

I don't think I'd predict that WebOS will be up near Android or the iPhone in 2012, but it'll likely still be around. The original analysis doesn't even mention them. Many people think that WebOS has the best technology of any of the existing mobile OSs, although they obviously need more apps (coming soon) and more phones (coming soon).

That's why you randomize, to avoid having to measure and control for the propensity to engage in risky behavior. I'm really puzzled what you think you're trying to argue. This is the simplest design in the world. Two equal groups of randomly-assigned people. One group gets a treatment, one gets a placebo. Neither group knows which group they're in. Measure the outcome. How would you do it better?

That's how random-assignment experimental design is done. It's not bogus. It's very well understood both statistically and philosophically. I'm sorry you don't like it, but it's been the basis of the scientific method for a couple hundred years...

Why would they take more risks? The whole point of a placebo is that you don't know if it's a placebo or not. So there's no reason to expect a change in behavior in one group versus the other. In fact, the behavior change should be driven in the other direction. If there was some reason to think that you got the vaccine (say, side effects not present with the placebo), then you would be likely to increase your risky behavior and increase your likelihood of infection! In this case, they got an effect in the other direction -- the treated group had less infections.

Not necessarily. HIV works by aikido-ing the immune system. It could be that the vaccine could either fully prevent infection or fully fail to prevent infection. Once the virus becomes established in immune cells, the presence of the vaccine and an immune response might be totally insignificant to the progress of the infection.

Yes, although there's an issue of multiple comparisons. There have been a fair number of HIV vaccine trials over the years. This is the first that's found statistically significant results. But if you were to test 20 different non-effective vaccines at a 5% significance level, you'd expect one of the tests to be significant just by chance. This is certainly an intriguing result, but it could be an outlier, and must be replicated.

I don't want have to decide *before* I read an article whether I want to pay for it, I want to decide *after*. To that end, I propose the following micropayment system. If I want to get content from a consortium of providers (say, anything owned by The New York Times Company, or Time-Warner, or Seed Media Group, or a group of publishers that set up their own consortium), I set up an account, pay my $50/year, and get access. If I like a piece of content (article, podcast, interactive graphic, whatever), I click the "Tip the Author(s)" button, and a chunk of my $50, maybe 10 cents, gets redirected to the actual people creating the content I actually like (not just start to read). If I don't use up my $50 for the year, it just gets split internally by the consortium. This way, readers have control over where the money goes and get to associate "paying money" with "feeling good about what they read", providers get cash, and the best providers get the most cash.