My views on value of infographics that actually look nice have changed, perhaps I should have had nicer looking figures.
“Unlike spreadsheets, users can describe in notebooks distributions, capturing uncertainty in their beliefs.” seems overconfident. Monte carlo has been an excel feature for years. My dismissal of this (implicit) makes sense as a thing to say because “which usecases are easy?” is a more important question than “which usecases can you get away with if you squint?”, but I could’ve done a way better job at actually reading and critiquing excel programs that feature monte carlo.
Viv told me to fix this and I ignored them because I liked the aesthetics of the sentence at the time: “which defines compositional systems as fluid de- and re-composition under the abolition of emergent properties”—viv was right. I’ve changed my views on the importance of writing being boring / non-idiosyncratic, also even my prose aesthetic preferences change over time (I no longer enjoy this sentence).
“(this use case isn’t totally unlike what Ergo accomplishes)” I keep thinking about ergo, perhaps a paragraph in the “whiggish history” section would’ve been appropriate, since API access to some zoo of scoring rules / crowd wisdom is a pretty obvious feature that many platforms would foreseeably prioritize.
I underrated how strong the claims about compositionality (being precisely the sum of parts), since statistical noise is such a fundamental piece of the puzzle.
I haven’t been working on this stuff except a little on the side for most of the last year, but still get excited here and there. I returned to this post because I might have another post about module systems in software design, package management, and estimational programming written up in the not too distant future.
the author re-reading one year+ out:
My views on value of infographics that actually look nice have changed, perhaps I should have had nicer looking figures.
“Unlike spreadsheets, users can describe in notebooks distributions, capturing uncertainty in their beliefs.” seems overconfident. Monte carlo has been an excel feature for years. My dismissal of this (implicit) makes sense as a thing to say because “which usecases are easy?” is a more important question than “which usecases can you get away with if you squint?”, but I could’ve done a way better job at actually reading and critiquing excel programs that feature monte carlo.
Viv told me to fix this and I ignored them because I liked the aesthetics of the sentence at the time: “which defines compositional systems as fluid de- and re-composition under the abolition of emergent properties”—viv was right. I’ve changed my views on the importance of writing being boring / non-idiosyncratic, also even my prose aesthetic preferences change over time (I no longer enjoy this sentence).
“(this use case isn’t totally unlike what Ergo accomplishes)” I keep thinking about ergo, perhaps a paragraph in the “whiggish history” section would’ve been appropriate, since API access to some zoo of scoring rules / crowd wisdom is a pretty obvious feature that many platforms would foreseeably prioritize.
I underrated how strong the claims about compositionality (being precisely the sum of parts), since statistical noise is such a fundamental piece of the puzzle.
I haven’t been working on this stuff except a little on the side for most of the last year, but still get excited here and there. I returned to this post because I might have another post about module systems in software design, package management, and estimational programming written up in the not too distant future.
Overall, this remains a super underrated area.