Daniel Lemire's blog

, 2 min read

When has a problem been solved?

2 thoughts on “When has a problem been solved?”

  1. Will says:

    Some notes:

    – I’d give a Nobel Prize to anyone who could reduce global warming/decrease cancer rates/improve worldwide health by 1%.

    – Spam will always have a precision/recall tradeoff, I think. I can stop your FP rate by blocking your access to mail.

    – Spammers will continue to improve, so an ‘order of magnitude’ improvement might just be to stay at a current annoyance level.

    – The definition of ‘spam’ will always fuzzy, as well.

    – I dislike your spam protection 🙂

  2. John Riedl says:

    I like Daniel’s high level perspective: researchers should focus on work that has room to have a big effect. Once we get down to the minor details it’s hard for researchers to have much of an impact, because the actual environment in which the research is deployed may matter more than the algorithm/interface/whatever discovered by the researcher.

    OTOH, I’m not sure there’s a “order of magnitude” definition that we’re all going to agree to. I think it’s very dependent on the domain, and the needs of users within that domain. To me the key question is: will this research, if it succeeds, change the user experience in a qualitative way? If so, I’m interested; if not, less so.