Hacker News Reader

Comment by sago | Hacker News Reader

I had dinner with Prof Minksy, about 15 years ago. He asked what I was working on (evolutionary algorithms) we got talking about domains of knowledge.

He said all knowledge has a half-life, the time it takes for half of what you know to be redundant or wrong. Math, he said, is the longest, measured in centuries or millennia. One should feel sorry for neuroscientists: they can go to the bathroom and half their knowledge will be out of date.

The only time I spent with him, but both the metaphor and his passion for math will stay with me.

View on HN

>neuroscientists: they can go to the bathroom and half their knowledge will be out of date

Web development seems to be even worse than neuroscience in this regard. I feel like I have to perpetually study to be abreast of best practices.

Well, at least the languages you use stay pretty much the same? Javascript, HTML, and CSS are always going to be there. Although.. I suppose that could be considered both good and bad..

While that's half-true, compare ES3 to ES2015, or DHTML to HTML5.

> Web development seems to be even worse than neuroscience in this regard. I feel like I have to perpetually study to be abreast of best practices.

The worst part is web development is a completely arbitrary made-up system without any abstractions you can learn and use in a different field of work. It does not even transfer to something as closely related as desktop/mobile GUI toolkits. I view "knowledge" like that as a waste of time and avoid it whenever possible. I wish there was a word for it, something to imply "garbage knowledge."

Really? Functional reactive programming, immutable data, unidirectional data flow, application state as a fold over a stream of actions... it seems like web development is seeing a really high uptake of valuable, transferable abstractions and ideas, far more than mobile GUI toolkits, for example.

JavaScript programming is, CSS and HTML not so much. I would also argue that the details of frameworks like Angular also fall in the "garbage knowledge" pile.

Fair enough, you meant to refer to a subset of web development, rather than the entire field. It's a significantly weaker claim, though. I could pick out subsets of any field of work, and declare them non-transferable, and the details of most systems are going to be less transferable than the concepts undergirding them.

That's a pretty interesting way to put it. I feel that the math-related fields also have a long half-life: EE, ME, Statistics, Physics, etc.

The EE and ME stuff won't exactly be wrong, but could become out of date because of new device types and materials. Imagine studying EE before solid-state devices and ICs, for example. But it's still a pretty long half-life compared to biology!

Good point. Still, a drastic revolution in devices isn't something that happens often.

This is super interesting. I tried to look for any reference of this, but found nothing. Do you happen to know if this thought of his is available somewhere, in a book, a paper, etc? Or is it a concept borrowed from someone else?

There is a wikipedia entry on the concept: https://en.wikipedia.org/wiki/Half-life_of_knowledge

Thanks. Useful.

Reply on news.ycombinator.com