It seems like a clear case where the law didn’t keep up with technology (although I wouldn’t expect it to). It’s difficult to conclude anything without understanding the actual law, but the gist is that email that’s been on a mail server for over 6 months is considered abandoned and therefore doesn’t require a warrant. It seems like the law took POP3 into account where we all download our mail, but not the modern world of leaving everything on IMAP and cloud storage.
According to the Wired article, the justice department wants to renew the act without fixing this part, because it would be “an unnecessary burden on the government”. I hope there was some misunderstanding, because this seems REALLY sketchy.
This week’s followup is that an English professor made 60,000 edits to Wikipedia and it helped in his tenure case. There’s more information over on the Wikimedia blog. It’s very interesting and he’s made quite a contribution to the general body of knowledge in that manner, but the articles don’t clarify whether it meaningfully strengthened his tenure case or not.
Would I do it? Maybe some articles, but not that level. Would I include it in a tenure case? Maybe in a service section, but definitely not in a research section. Do I think I’d be tenured with Wikipedia contributions and not tenured without? No. The bottom line for teaching schools is teaching (and research). The bottom line for research schools is research. Even though it’s a meaningful contribution to society, it’d have to be a really borderline case to make a difference.
Instead, I view blogging as part of my contribution to society. In this case, I can write almost anything I’d like without fear of editing, and my work is properly attributed. Wikipedia just isn’t the place for certain content (like say advice for grad school). Other content (like a paper review) is too specific for Wikipedia. Also, in a sense blogging improves my “personal brand” whereas Wikipedia editing doesn’t seem to.
Today’s xkcd comic illustrates a problem in designing subjective user evaluations. In this case, the experimenter tries to use an example to explain what the 1-10 scale represents. The goal of the experimenter is to normalize the scale across people – we want responses of “1” to mean the same for everyone, “5”, “10”, etc. But because the reference points weren’t specific enough, the researcher introduced a significant source of variation into their study. You may laugh because it’s a comic, but it’s a real-world issue in user studies.
Wednesday’s xkcd brings up an interesting statistical problem – the more significance tests you run at some cutoff, the greater chance that one of them happens to be significant (due to chance). Language Log does a followup about accounting for this increased likelihood that one test happens to be significant using Bonferroni correction.
Stepping back for a moment, there are two issues this reminds me of. At the bottom of the comic, it gives an example of how this error could lead to silly news stories. I’d suggest that research reporting requires some degree of statistical and scientific expertise.
The other thing is that I never knew about Bonferroni correction until a friend of mine mentioned it. The only formal statistics training I had was a single class in undergrad. Reflecting on that a bit, it’s partly my own fault for lack of foresight in taking classes (both in undergrad and grad school). And it’s partly a failing of grad programs – we’re expected to produce quality studies, writing, and teaching but yet we’re not required to learn about any of that.