There's such a huge difference between reading a paper about an algorithm/system/etc., and trying to implement it yourself.

I read the paper and think "I get it".

I try to implement it and realize I didn't get it *at all*.

(This tweet brought to you by Concurrent ML.)

Context: I'm still exploring models for concurrency as discussed in my last blog post. Got turned on to CML by @andywingo's excellent articles about it, e.g. wingolog.org/archives/2017/….
twitter.com/dubroy/status/…

@andywingo Related — @danluu on noticing "the feeling of glossing over something without really understanding it".

danluu.com/cocktail-ideas/