Hacker News Reader

I like the reach for Amdahl's Law, but it's about how fast you can speed something up by doing it in parallel, isn't it? What's to say someone else can't be eliminating the other classes of homelessness in parallel, making the problem effectively embarrassingly parallelizable?

View on HN

> What's to say someone else can't be eliminating the other classes of homelessness in parallel

You both can and must address everything in parallel to actually dent the "homeless problem".

That's the point.

Let's say there are 5 primary classes of homeless. Even if you take one of them and reduce it to zero (and you will never accomplish this)--you're still left with 80% of the problem.

Only now the people approving the checks are angry because the "homeless problem" doesn't really look any better.

Hmm. My takeaway from Amdahl's Law is that if one has a parallelizable problem, one can keep throwing processors at it, but eventually there is a slice of overall work that cannot be parallelized. Stepping back and refreshing myself on Amdahl's Law, I can see how this relates to Amdahl's Law. I like it.

Anyway, I agree with your general sentiment, and it tickles my mind. Thank you!

Reply on news.ycombinator.com