Thursday, July 27, 2006

The Collected Works of Lance Fortnow

In the ancient days of the early '80s when you wanted a copy of a research paper and didn't have it in your library, you sent a stamped self-addressed envelope to one of the authors who would send the paper back to you.

I started graduate school in 1985 as a member of that new-fangled email generation. When I got an email request for a paper, I tried sending a LaTeX file, sometimes getting the response "What am I supposed to do with this? Can't you just send me the paper in the mail?"

But as people became more comfortable with email I got many more email requests for papers, and responding to those requests took some time. When the the web started in the early '90s I set up a page for people to download the papers. I would still get email requests for a while, responding to the request but with a reminder that they could download my papers online. Around 1994, there was a phase shift and I nearly stopped getting any email requests, everyone knew to look for downloads first. Now most researchers put copies of their papers online and shame on those that don't.

I use a now ancient version of bib2html to generate the publications page. It makes for a functional but not very pretty page. I use the same bib file to generate both the webpage and the papers list in my CV.

I tell this story because I made the first major change on my publications page in about ten years. It looks pretty much the same, but I added links in the titles to the official publisher's page for the paper. These pages often give an abstract and if you have permission you can download the "official" version of a paper. If you don't have permission you can still download the unofficial version from my page. The publisher's page also gives you a way to link to an official description of the paper without linking directly to a file, something I like to do when I link to papers in this weblog.

I tried to use DOIs when possible or other permanent links so the links shouldn't go bad. The IEEE and the IEEE Computer Society (publishers of the FOCS and Complexity proceedings) maintain two separate digital libraries with some overlap. When I had the choice I linked to the IEEE-CS version because at the University of Chicago we have access to the IEEE-CS downloads but not the general IEEE.

Many other researchers, for example Salil Vadhan, do far better than me, maintaining their own separate page for each paper and sorting their papers by research area. Those young scientists always showing up their elders.


  1. Seems like finding tools to automatically generate a web page of publications to their preference is a problem that every researcher (re)solves for themself -- especially given that there are multiple different pieces of software that all use the name bib2html or bibtex2html. Do most people make their pages by hand, using a bibtex converter directly, or post-process the results of a conversion with additional scripts (as I do), I wonder?


  2. Lance don't take this personally,
    it's only a constructive comment
    that I hope you will find useful.

    About your publications web
    page, new and old, I always found funny
    that the "Important Notes" link at the top
    pointing to some copyright disclaimers,
    looks very much like the title of a
    section. So the very first time I saw it I
    thought: "Strange. A reasonably modest
    person wouldn't refer to his own
    publications as 'Important Notes'" :)

    Not that they aren't, but...

  3. Lance, Just wanted to say that I like this idea of linking to the official publisher's page on top of providing the files directly. If I ever do a significant revamp of my page at some point, I might try and do the same.

  4. If a paper is on ECCC or on the arxiv, then a link to there goes to a page that contains an abstract as well as a version of the paper that everybody can download.

  5. Seriously, though:

    <META NAME="GENERATOR" CONTENT="Internet Assistant for Word 1.0Z">