Tikhon Jelvis

@tikhonjelvis

I like programming languages. A lot. Especially Haskell. Principal AI Scientist at Target

Berkeley, CA
Joined September 2010

Tweets

You blocked @tikhonjelvis

Are you sure you want to view these Tweets? Viewing Tweets won't unblock @tikhonjelvis

  1. Making important tasks like writing tests *less boring* would go a long way to improving software engineering quality in the real world—but I don't see people talking about that very much :P.

    Undo
  2. Retweeted
    Replying to

    I'll do my best. Here's my story. I was asked to teach upper-level software engineering many years ago. How hard can it be, I thought. I do SE, I publish in all the major SE confs, this should be easy. I got about a dozen SE books to read over winter break to design my course. »

    Undo
  3. Retweeted
    Jan 29

    What should be humbling about this for CS educators is that in CS classrooms, DSLs are still thought to be esoteric and fancy. But to paraphrase Ivan Sutherland, the community as a whole doesn’t know this is supposed to be A Hard Thing, so they just do it! Incredible.

    Show this thread
    Undo
  4. Jan 29

    ...and people tell me Haskell is too abstract.

    Undo
  5. Jan 29

    Everyone should settle on a single framework for machine learning. Then they could call it Standard ML.

    Undo
  6. Jan 26

    Unstructured interviews simply don't work. People who think "I can just talk to a candidate and know how good they are" are misleading themselves, but, man, is it hard to convince them of that... Anyway, here's a solid overview on the subject:

    Undo
  7. Retweeted
    Jan 24

    One assumption that underlies the belief that Jira is good is that a backlog is an infinitely long list of up-front requirements captured by experts who pass those requirements to teams to implement as specified. 1/6

    Show this thread
    Undo
  8. Retweeted
    Jan 24

    Brilliant! Though the problem with new grads is that most haven't done anything and have the same 3 class projects. That and resumes have been shown to not predict on-the-job success. Everyone in the eng world should just use and be done with it amirite.

    Undo
  9. Jan 22

    Imagine a committee with veto power over aesthetic decisions, made up of the sort of people who get actively involved in municipal politics and probably run your local HOA. How could anything *not* bland make it through?

    Show this thread
    Undo
  10. Jan 22

    Makes me feel better about shipping buggy software cause at worst it'll make Target's supply chain less efficient and not, you know, nuke Arkansas.

    Show this thread
    Undo
  11. Jan 22

    Now that I've finished Command and Control, it's my first 5 star book of the year. A surprisingly intense history of the (lack of) safety in the US's nuclear weapons program, it covered a lot of ground while still keeping my attention.

    Show this thread
    Undo
  12. Jan 21

    The expressiveness of a programming language is one of those things that I think is important, but I can't define it quantify. It's a frustrating "I know it when I see it" situation, which is such a weasely phrase.

    Undo
  13. Jan 21

    I *have* written a bunch of Elisp in the interim, and that's basically the same thing, right? I mean, what's a bit of dynamic scope between friends?

    Show this thread
    Undo
  14. Jan 21

    Just got my ticket. I haven't used Racket in a while, so it'll be cool to see what's up in the Racket world.

    Show this thread
    Undo
  15. Jan 20

    It's a great followup to The Best and the Brightest which I read last year. That book focuses on how the US government got itself into the Vietnam War—which happened around the same period with the same high-level leaders as Command and Control covers.

    Show this thread
    Undo
  16. Jan 20
    Show this thread
    Undo
  17. Jan 20

    I'm reading Command and Control—a fascinating and disturbing account of how the government handled nuclear weapons safety historically. Great reminder that large organizations can be stunningly incompetent no matter how high the stakes or how smart the individuals involved.

    Show this thread
    Undo
  18. Jan 20

    And it might be really difficult to fix an architectural problem incrementally. What should you do? I don't really know myself. Just try not to get into that situation in the first place!

    Show this thread
    Undo
  19. Jan 20

    Concrete example from my own (painful) experience: if the architecture of your code forces every feature to touch thousands of lines of existing code, that problem won't be affected much no matter how you improve your tests, documentation... etc.

    Show this thread
    Undo

Loading seems to be taking a while.

Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.

    You may also like

    ·