Joel Spolsky says some interesting stuff about software at joelonsoftware.com.
The New York Review of Books has wonderful writing on all sorts of things at nyrb.com.
I recently came across Charles Petzold's page at charlespetzold.com.
TopCoder.com has, among other things, a good programming problem archive to keep your brain healthy.
Paul Graham has some interesting commentary on startups, ideas, hacking, programming, and other topics at paulgraham.com.
I will be a graduate student in a year or two. In the meantime, I try to keep up with research papers when I have time. Here are a few that I think are either seminal, interesting, or just cool.
I just came across this one recently. Databases are interesting to me because they are a whole bunch of gnarly computer science problems in one place. This is a paper about C-Store, a database that stores columns contiguously instead of rows. I think it's cool because it's a new take at an old technology. It's also a problem I've been working on.
In particular, I like that the authors address the huge problem of having to have two databases: one for OLTP and another for OLAP.
The Naive Bayes Classifier is really cool because it's such a simple thing but it works quite well. This is a paper that discusses how good the NBC is and explains why it probably shouldn't be.
This paper highlights a few well-known problems that exhibit a "phase shift" at certain points from being hard problems to being really hard problems.
This is the "Google Paper", one of the early papers about how PageRank works. This is a seminal paper because, as everyone knows by now, there were lots of other companies in the search engine business before Google came along, but shortly thereafter they disappeared. Brin and Page took a look at how search engines worked and decided there was a better way. This takes some guts, and I like that. This sort of thing doesn't happen often enough, IMO.
Here is a great paper that explains the math of information retrieval. Long, but good.
Latent Semantic Indexing is the coolest IR technique that I've seen. It is described in the previous paper about the math of IR, but it was first described by Scott Deerwester before IR was a huge field. Here's that paper.