Jaime Brandon Call 9/5/17

• part time master in cognitive science at UCL

• replication crisis in softer sciences in last decade or so, lsat 40-50 years, subtle effects of statistics. TODO 7 deadly sins of psychology http://scattered-thoughts.net/blog/2017/05/18/notes-on-the-seven-deadly-sins-of-psychology-a-manifesto-for-reforming-the-culture-of-scientific-practice/

• check out http://tiddlywiki.com/ for better system than github pages. has some neat ideas, but feels like it needs to be unified more

• in order to understand query, datalog, read about logicblox, like https://pdfs.semanticscholar.org/0ffa/f68f4399998864d6d9835c7bd8240d322b49.pdf

• currently being paid to work 3 days a week to work on whatever I feel like… supported by logicblox. got them by first building a query compiler first and then showing it around. then the first company that he went to, logicblox, said yes. he was prepared to go around to multiple companies but the first said yes. He mostly talked with the CEO, but he just quit so now the CEO is supporting Jaime personally. Maybe Jaime would start a company with him when he’s ready, after his vacation. This is exciting to Jaime because they had been successful in the past, from an execution perspective. They had a buisiness that got money and shipped tech.

• TODO send Jaime Paul Biggar Dark stuff

• StreamSheets feedback
  • it is basically “event sourcing”, but I have to be careful that there are no side-effects by accident or you lose a lot of the benefits of it
  • Google: Frank McSharey Timely dataflow
  • MIT Objective spreadsheets https://sdg.csail.mit.edu/projects/objsheets
  • “the spreadsheet is the universial UI for timeseries data”

• TODO think about what it would look like to have a branch structure where it encourages you to create smaller and smaller features and merge them early

• Next few weeks for him: trying to keep things small and self contained so he can actually accomplish things. Could be a good platform for keeping track of what paper you’ve read, want to read, notes you’ve taken on. Part-database and part-wiki.

• Making each commit and feature really small
  • What is that called? “accereting a system instead of desigining it”
  • once you come to a feature you want, you first have to list all the possible ways to get it
    • building it custom yourself
    • tailoring something that exsits to your use
  • you have to think about how far away each of the “comprimises” you choose will take you from your vision
  • In this conversation, we chatted about how JS and shipping SQL to SQlite on the backend via sending SQL is the quickest way. What’s the tradeoff between doing this and another option?
• LogicHub feedback
  • https://blog.ought.com/dialog-markets-37db7c185e41
  • Made Jaime realize that if there was a tool that made him justify why he wasn’t using JS and SQLite, that would help him realize that the reason he isn’t is because they are “icky” in his head and when he reflects on why he’s not using it, he just thinks “there must be some reason”. TODO think about whether it could somehow be integrated with WoofJS worklow in ways
  • He has a decision template and heuristics but always forgets to use it in the moment and is wondering how to make sure that he is reminded through triggers and then checks

• TODO think about the recurring reminders I want to think about and make sure I check regularly

• software knowledge transfer of best practices doesn’t happen as much (like me seeing Jaime’s journal) in the same way as you would if you were a mathmatician and could watch a peer do a proof

• TODO add Jaime to slack group

• how they run liberal media stuff
  • 15 min of saying hi
  • then pair people randomly and send them on 15min walks (TODO add this to list of options for future programming meetings)