[split] MFMP experiments...

    • Official Post

    Where you just talk about MFMP: What's their state? Are they doing experiments? They have laaarge down-times between each test.


    I guess @me356 approach is much more effective. Run 5-10 experients with small differences in parallel. Chose the most promising and develop 10 new experiments using the design from the last iteration.
    It's much work, but very effective. Like DNA evolution. Billions of modifications are "tested" in parallel, day by day.


    I guess this way he found his "secret".

  • I'm still a bit confused about the lack of summary clarity on the MFMP site. It may be just because I'm not reading it enough, but from the outside there work would be more accessible if given say a 6 month "state of the mission" summary. With care that could be both accurate and positive because getting more robust experiments, even when results are not positive, is worthwhile. It is easy to criticise such things from the outside of course.


    I have found the MFMP site and reports to be frustratingly difficult to access, so I stopped watching. The potential value of MFMP work is very high, but they will need communicators to fully realize it. Are they seeking them?


    These would be people with a general understanding of the issues and the work, perhaps, but most critically, people who would communicate with the actual workers and report on projects and progress, maintaining summaries and presentation of results to date.


    Are they doing this? I don't know. They lost my attention some time back.

  • In reply to the several comments above on MFMP's work, here's a document summarizing our recent experiments and ongoing plans. The content is based on a talk I just gave at the Anthropocene Institute in Menlo Park (Silicon Valley). Please post or quote this document ONLY by reference to the link below, so that ongoing revisions and additions will propagate to all online copies.


    http://goo.gl/Zo0Yr9


    Alan Goldwater
    MFMP