Log in

No account? Create an account
color cycle (slow)

Kistaro Windrider, Reptillian Situation Assessor

Unfortunately, I Really Am That Nerdy

Previous Entry Share Next Entry
A Decidedly Nifty Day
I've been silent on this journal and difficult to reach online the last few days because work has kept me really busy- I've been working on my laptop at my apartment after hours rather regularly the last few nights. I had to get the project complete, because there's little time left- these ten and a half weeks so far have just flown by. This is the middle of my second-to-last week, and much of next week will be preparing my final presentation and preparing for my final interviews. Today was my major presentation of what I did to my team, with full information on how they are supposed to use the API I wrote.

Somehow, I took four conflicting sets of requirements that kept changing and made something that everybody is happy with.

My supervisor told me he hopes to see me in a year, in a full-time position.

He has the strongest vote on whether or not I will be given a full-time offer; because he has more direct experience with me, he even gets a stronger hand in the decision than the division manager; his assessment will be considered in more detail than that of the interviewer I'll meet the day before my last day. And because my mentor feels the same way, as do my co-workers, and I know I did a lot for three months, I think my odds are really good that I'll get that offer...

The other nifty thing: http://stardustathome.ssl.berkeley.edu is a web site attached to a major research project analyzing aerogel collected by the NASA "Stardust" probe, which had one block of aerogel exposed to the dust of a comet and one that was exposed just to general interstellar dust. The first is easy to find. The second? Well, they're expecting somewhere around 45 particles of dust, about one micron across each, within a 10m-square block of aerogel. So they took lots and lots and lots of microscope pictures of the aerogel at different focuses, and posted them on the Internet there. The first step is to find the surface; the second step is to focus below the surface to look for trails of particles. And there are lots and lots of things to look through... which is where everybody's help comes in. It's a distributed project of many eyes, relying on Internet users going through a short training course and then passing an entrance exam to identify which blocks have traces that might be particles and which ones can be ignored- and which ones never show the right focus and need to be re-scanned.

To keep you on your toes, about 1/5 of the trials aren't from Stardust, they're lab examples, or they are from Stardust but they've already been professionally analyzed. You get a score for using the site: for each of these "calibration trials" that aren't a real unknown, you get a point for each one you get right and lose a point for each one you get wrong. Your score isn't just for bragging: the higher your score and the better your hit percentage (percentage is more important than net score), the higher priority you get. The higher your priority, the higher samples you mark as "there's a trail here" will be placed in the queue for the scientists to analyze. Samples are given to more than one person, and multiple people agreeing that there's a trail wll get a region kicked to the top of the queue.

So far, in one evening, I've scored 40 points for 40 calibration hits and no misses; my score is a double 100%. By accuracy, I'm tied for top (perfect). By score, I'm in 435th place of 3648 users who passed the qualification test. It's pretty cool to have a good rank like that, but 3648 isn't nearly enough people. But it just hit Slashdot, so that might be going up...

I guess I'm mostly advertising it, because it is so much nifty and y'all should try it. Seriously. Maybe you'll discover something useful...

It was an immensely gratifying moment when I looked at a set of scans, saw something, checked it again, said to myself "Oh, this one must be a lab example, it's too clear", and marked the spot. (Marking the spot may require a double-click, not a single-click as the instructions imply.) Of course, it was a real trial, and I'm not the only one who saw something in it... There are three other interesting but subtle traces I found that nobody else agrees with me on, but maybe they'll be analyzed anyway on the strength of my high rating.

It's a very fascinating way to spend time. And it's just a generally cool thing to do.

  • 1
Congratulations on your success at MS. I'm sure it will give you a leg up whether you go on to work there or go somewhere else.

Hmmm, the stardust project is using peer calibrated review. That's perfect; I use something like that to evaluate projects that I make my Calc II students write. They bitch about doing projects, but having students complain about having to do more than sit there is bound to set someone off. They also complain about how they get grades:

I give them five projects from previous semesters, including two that I purposefully wrote very poorly--good math, confusing writing, with one major error that I just ignore and somehow get the right answer. They're supposed to rank the five projects from best to worst.

Then when they write their own, I strip off their names and make copies and give them back out, with the same instructions. They are often much more exacting and picky than I would ever be when they grade each other's work, that's the only drawback.

  • 1