This Week's Higgledy-piggledy

Links, bits, bobs

Marsrover, getting to know Marseille

Startup world

Other stuff

Software development

Random thoughts

On starting a startup

Coming back on the Sociology of Steve Jobs linked to last week: you need deep confidence to sustain the level of action and detached evaluation of the world that Jobs seemed to have. That comes through frequent external feedback, be it positive or negative.

Why do people start startups? I wouldn't do it only for the money, it's easier to work in a big company for that. I believe a job makes sense if it has at least two of the following:

  • it's an activity you like,
  • it has meaning,
  • or it pays.

Now meaning comes from things you believe in, your life environment and your community. Something you want to be part of, you want to create, or you want to change.

Because that's the thing with tech, there's this incredible hegemony of people pontificating on how things are different in that area, but should be done this or that way. And they're very vocal. But rarely do you read a piece that isn't tainted by a conflict of interest (e.g. people explaining why this or that is good, when they're really just defending the choices they made).

Programming languages

Idea: evolve a programming language on Mechanical Turk by evolving features with user tests (thought from Maybe I Actually Am An Engineer).

On problems and solutions

A lot of things I read (a lot of things published?) are solutions, not problems. Sometimes solutions to no problems. But it's way more empowering to read about problems that seem solvable, but not solved (or with no proposed solution).

Problems I find interesting:

  • academic search (for papers, for people) and evaluation is broken. Asking for more money, or reducing the problem to lack of money, is not an excuse: there will never be enough money, and the competitiveness pressure will never disappear, so the problem must be solved as is. Solution: many possible things! Ideas: start with search, then expand to publication format, then enactive communication devices.
  • Silicon Valley is a monoculture in a single geographical spot. Solution: oh god, let's brainstorm.
  • University is broken. It's being eaten by budget cuts, it's being eaten by shorter thought cycles and shorter project cycles (and maybe, in fact, the whole project-based thinking).

Random ideas

  • Make a browser extension that shows you the star growth rate of GitHub repositories.


I read Rodney Brooks' Elephants Don't Play Chess this week, and mother of god! What the heck happened after this? How come, over 25 years later, we don't have a swarm of autonomous robots in our homes and in the streets (e.g. picking up trash)? Not even a single one? (Apart from people with robot hoovers, which seems to be the only civilian industrial application that came out of that research.)

I was honestly surprised at the level of autonomy the robots described in that paper have, and I consider myself relatively well informed on the subject. Are the only outcomes of this robot hoovers and self-driving cars? Did the forget-ideas-and-cement-current-practice phenomenon happen in AI research the same way it happened in programming?

And that's enough for this week!