The Mist

The Mist is a movie, adapted from a Stephen King novel. It is a bad movie. Don’t see it.

Actually, I realized after watching it that The Mist may be a genuine modern (2007) B Movie. It’s a classic monster movie, complete with unconvincing tentacles, ridiculously stereotyped characters, lines so terrible that even the relatively famous actors can’t deliver them, and a mysteriously buff, resolutely honorable protagonist who, just to make sure we understand that this is fiction, pays the bills on his gorgeous Victorian lakefront house by painting movie posters. There’s also a tremendous amount of screaming, none of it believable.

On the other hand, if you’re in the mood for a senseless, predictable, self-serious movie about gigantic invading alien bugs, then you should see it.


When I got home today, I tried to connect to the wireless network. The connection wasn’t happy, so I took a look at our wireless router, which is also the DSL modem. It was blinking in an unhappy-looking way, so I pulled the power and reset it. This didn’t improve the situation, so I sat down at the desktop, which has a direct ethernet connection, and attempted to connect there. No luck. I did manage to bring up the router’s configuration console, but it had been replaced by a simple webpage saying something like “The router is not functioning. Either a firmware upgrade has failed to complete properly, or there is another source of memory corruption. Please contact Technical Support.” After a lot of plugging and unplugging while talking to technical support, we concluded that the router was unrepairable.

My dad is suspicious of power electronics, so we checked the voltage on the power brick. It was actually slightly above its nominal 6VDC. He still suspected that the onboard step-down DC-DC converter may have been getting weaker over time, leading to system failure once the main components kick in, but there’s no way to test that.

But I just made a startling discovery: stopping my laptop’s wireless ethernet system makes the router start working again, and turning my laptop’s wireless ethernet back on kills the router again. This is a serious incompatibility; my laptop is effectively hacking into the router and disabling it, accidentally. It’s not just any wireless device that does this. My cell phone, which also uses wireless ethernet, connects just fine.

I’m not sure what I’ll do next. File a bug report with the people who wrote my driver, I guess. At least we don’t have to buy a new router.

Edit: The people who wrote my driver are Intel, and they already know.

Oil Change

I’m getting my oil changed at the dealership. I am waiting in the lobby while they do the work (it’s not like I have anywhere else to go). On the 40″ plasma screen at the front of the waiting room they are showing NBC.

Normally, this would be fine, but NBC at 1 PM on a Friday apparently means “Days of Our Lives”, an incredibly trashy, annoying soap opera that is currently showing some kind of weird soap opera sex scene.



Saturday, I went to my boss/coworker/labmate/advisor’s house for a Fondue party. It was wonderful, and tasty, and I am kind of scared to think about how much cheese I must have eaten.

Sunday, a friend held a “Easter Keg Hunt”, which involves Easter, and a Keg, but not really any Hunting. That was also fantastic.

Today, we started another overnight scan. I would cross my fingers and hope that it works, but I’m typing.

It’s spring break. Life is good.


It’s spring break for Harvard, which basically means I get to spend more time in lab. I ran to the gym today, and when I got there, it was locked. The sign on the door said “closed for spring break”, so I ran back home.

Hey Harvard, did you ever think that maybe students would have more free time over spring break, and so they might be more likely to use the athletic facilities?



I’m famous! OK, I’m not mentioned by name. Somebody’s famous, and that somebody is me. And I’m famous for writing inscrutable software… whatever.

Tonight was a wine & cheese party. It was fun. We talked about AI and quantum theory and NP-Hard artificial economies, and all the usual things you’d expect me to talk about, only there were other people there who were also talking about them. Super-cool.

In keeping with the theme, I brought a brick of Emmenthaler, which I thought I remembered as my new favorite cheese. Unfortunately my memory must be faulty, because this seems not to be that totally awesome cheese I’ve been having at Schusheim family reunions. Hey mom, what’s that cheese called?

Flash is Cheap

If you go here, you can buy a 4GB USB flash memory drive for $16. That’s the same amount of data storage as a typical DVD, but infinitely reusable, and in a package much smaller than, say, a pencil. It also costs less than buying a DVD. Want 8 GB? That’s $30, for the same physical size. Since they don’t have any moving parts or exposed fragile bits, they’re likely to last for decades. It makes one want to reconsider ever using DVDs again.

The cost works out to about $4/GB for flash, or a little less. Adjusted for inflation, that’s about same cost as spinning hard drives in 2002, according to this site. Why are we still using fragile, bulky, unreliable spinning disks at all? Because spinning disk today is down to about $0.16/GB. That’s a factor of 25 in price over the past 6 years, or a price-halving time of about 15 months. Hard drive capacities have hardly grown over the past 5 years, but for large capacities, the $/GB is becoming very low indeed.

I’ve often wondered how Google could possibly afford to offer 6 GB of storage to every single GMail user, for free. One presumes that they do not actually have that much storage anyway, but it still seemed like a lot. Now I know the answer: at today’s prices, 6 GB is $1. If every single GMail user unexpectedly maxed out their inbox capacity, Google could buy the required storage in cash without a second thought.

Also, a good way of thinking about this is to imagine asking every human to remember one letter. The amount of data you can store that way is 6 GB, which costs about $1 these days. For $1000, you can buy enough storage for each person to write a sentence. With a $1 million worth of storage (fairly common in large datacenters) you have enough room for every person on earth to write their own 500 page book (actually more like 5000 pages with simple efficiency tricks).


The Boston T has replaced tokens with contactless payment passes that may simply be tapped on a sensor at the gate in order to debit the fare ($1.70) and open the gate. I bought my first card in December, I think, once I’d started taking the T regularly to lab. Every time my card ran too low, I added $5 to it. This weekend, I hit $0.00 for the first time. How many T fares had I taken?

We are looking for some integer n such that n * 170 = 0 mod 500. The least such n is 50, indicating 50 T rides since December, or $85 spent on the T. This validates my decision not to buy a monthly pass, which would have been somewhat more expensive. On the other hand, it also suggests that I should probably be adding more than $5 to my card.

Recruiting over

Today was the last day of Biophysics recruiting. In the morning was breakfast, then three lectures by faculty, then lunch, a poster session, dinner, and a field trip to “Avenue Q”.

Every meal was a feast. I think that means I’ve been to at least 6 Biophysics-sponsored feasts, and possibly 7, counting the wonderful “business lunch” we had for MCB199 staff at that Chinese buffet. The students were very cool; I think we’ll be seeing a lot of them again.

“Avenue Q” is a new Broadway show. It’s a comedy that draws its humor primarily from shock value and abuse of the Sesame Street style, which it replicates in every detail. If I’d been a more ardent fan of Sesame Street, or if I hadn’t already known all the songs, I would have liked it more. As it was, I was not especially impressed. If you’re planning to go, I recommend you not listen to the soundtrack ahead of time. The jokes are funny the first time.


It’s been a very busy week. The Biophysics Department’s recruiting “weekend” started Wednesday night, with dinner at Om, an asian-themed overhip restaurant/lounge. Yesterday it was dinner at a fancy restaurant in the North End; tonight is the big ceremonial dinner thing.

Last night, the other TA and I attempted to grade the exam. It was hard, and around 2 AM we gave up, and I crashed on his couch. This morning I performed final debug on a pulse sequence that’s set to run at 2 PM. At that point, I’ll probably still be talking about these tests, trying to puzzle out how to grade people who took a completely different interpretation of the question and proceeded to answer it correctly.

Unrelated: I have an observation about the mortgage problem, and science. Prof. George Church, a genomicist (I made that word up), is doing a study in which he acquires complete genomes of about a dozen people, each one independent. He will then publish these genomes, along with a full biological parameterization of the individuals, so that other scientists can look for connections between the genes and the resulting human.

This is considered a highly risky human-subject experiment, because it will be relatively easy to figure out who the subjects are, even if their names are not given, because each person’s DNA is unique. There are all sorts of privacy implications, most notably the possibility that insurance companies could deny coverage on the basis that they know an individual is at high risk of particular diseases. Normally, no university ethics board would allow this experiment to go forward.

The Church lab has figured out a clever solution to this problem. In every human trial, the subjects are required to give “informed consent”. The scientists must provide them with an intelligible overview of all the risks associated with the study, and the subjects must understand this list and accept it. The Church lab, for this experiment, has created “highly informed consent”: because the risks might be considered especially high, and difficult to communicate well, only subjects with a “Master’s degree in genetics or equivalent” will be permitted to sign up for the study.

The news media tells me that millions of Americans recently signed contracts for adjustable-rate mortgages. These mortgages have unlimited downside risk, and are really quite dangerous financial instruments, like an uncovered call option. There are plenty of people for whom an adjustable-rate mortgage really was the correct choice, but I suspect many of the people who bought them did not correctly apprehend the dangers of this form. Most theories of contract law require a “meeting of minds”, in which both sides read the same contract, both fully understand its implications, and both agree to be bound by it. In these theories, a signed contract is not binding unless the signing party correctly understood its meaning. This is extremely similar to informed consent.

Congress is currently debating placing restrictions on what kinds of mortgages may legally be offered. This is not, I think, the best solution. The best solution is something like what the Church lab ended up with. If a party to a contract is exposed to unlimited downside risk (or appropriately defined “high risk”) then that party should have to demonstrate sufficient expertise in finance and financial history to prove that they fully understand the possible consequences of the contract they are signing. In my opinion, a solution of this form would be more effective, both in preserving the power of contracts and in preventing future financial excesses.