January 28, 2008
A New Term: "Dust Network"I saw a new term today (OK, it was last week, but I haven't had a chance to blog it). It's "dust network".
What is a dust network? The term is, amazingly, so new that there is no Wikipedia entry for it. Investigation reveals that it could be several things. First of all there is a company, Dust Networks, which makes embedded wireless sensor networking technology. That seems to be the gist of its genericized use also, in the sense of a bunch of very very cheap devices that communicate via a simple data stream to allow themselves to act as one big thing (the name "weak nuclear force interaction network" may give a better sense of what is meant, but doesn't roll as trippingly off the tongue). There is some notion that the data only flows from the devices, requiring a central computer to interpret the result and turn the "dust particles" into one very large and complicated sensor (here is a less-than-helpful page that asks the questions "What is a dust network?", but neglects to answer it). A company called Inrix (evidently a Microsoft spinoff) has a Smart Dust Network which integrates a bunch of sensors that can report information about traffic flow in order to provide a more accurate picture of current conditions. And here's a book called Smart Dust, which calls them "dust sensor networks". The phrase "smart dust" does appear in Wikipedia, referring to devices so small and simple that they can only detect light or vibration.
Dust network...remember, you heard it here first.
January 24, 2008
Yes, I'm Still AroundJust been a bit busy at work and outside of work, but I'll eventually blog something new.
But this isn't it.
January 13, 2008
Indicating Progress During PresentationsIn my job in Engineering Excellence, I occasionally put together slide presentations. I think it's important in a presentation to give the audience some notion of how far through you are. But I hate those intertitle slides that show "these are the five things we are going to talk about, and the current one is highlighted." So I try to be a bit more creative than that.
This is a talk I gave in 2006 about interviewing. The talk was split into 3 sections. In the intro slide I assigned each of them a shape (circle, square, and triangle). Then at the bottom I put three lines, each one representing the relative length of the sections (Cranky Kong appears in a few slides to remind me to tell a story):
Sp here we are starting out, we are in the first section which is denoted by a circle, and it's the first slide of that section so the circle is all the way to the left of the line:
Now we're near the beginning of the second section (there's Cranky reminding me to tell a story about an interview candidate who didn't know how many days there were in each month):
And then here is a slide in the middle of the last section:
Here's another example. I gave a talk last month about code reading, and to give it a seasonal gloss I decided to use a melting snowman to indicate progress. So here at the beginning is the full snowman (which originated as Microsoft clip art):
Now he is starting to melt:
Towards the end he is hardly there:
And (my final flourish) at the last slide, all that is left is his hat:
I really like the effect of both of these, especially the snowman if you flip through the slides quickly. The big problem is that if you add or remove a slide, you mess everything up. The first one, with the symbols moving over the bars, isn't that hard to do. But the snowman one took me about half an hour. I had to split the clip art into the snowman and the hat (since the hat didn't shrink), then shrink the snowman part by a precise amount each time, then move the snowman and the hat down (since when you scale something down it keeps the top border at the same spot, but I needed the bottom border to stay put). Hopefully at some point Powerpoint will add support for this kind of thing, since it wouldn't be that hard to do and then of course it could automatically adjust for slide counts.
Incidentally, if you work at Microsoft and you want to see the talks, they are both on mylearning. The first one is called "Choosing Technical Interview Questions" and the second is called "How to Read Code for Fun and Profit".
January 04, 2008
Checklists Get Nixed by the GummintWhen I read Atul Gawande's New Yorker article about how using checklists in hospitals can save lives, I had two thoughts:
- I hope if I wind up in the hospital, it's in one of the ones in Michigan that is using this technique.
- I hope this article will inspire/shame other hospitals into adopting it.
What actually happened, as Gawande explained in an op-ed piece in the New York Times last week, is that the government Office of Human Research Protections ordered the program shut down, because it was considered an experiment that had been done on patients without their consent (Gawande describes this as "blinkered logic". I can only dream that one day I will be a good enough writer to produce so apt an adjective in such a situation).
It would be quite sad if the New Yorker article was what alerted the authorities to the program and therefore led to the ban, but I doubt that government works that quickly (insert snide comment about readership of the New Yorker among Bush administration bureaucrats). The whole thing is quite depressing, because it seems so misguided. Gawande's argument is that this is not an experiment like a new drug; it's an experiment in healthcare delivery, which is different. He explains that this kind of work had previously been exempted from scientific research regulations, but now that it is becoming more scientific, ironically, it is being controlled. He points out that Congress could pass a law allowing it; of course this won't help the thousand people who will die next year because of the ruling.