Comment by DarkShikari | Hacker News Reader

Comment by DarkShikari | Hacker News Reader

Parent Full thread

I agree with the article for the most part. People like to think they're special because of their ideas, but ideas are easy to come by.

I find this strongly depends. I agree with the article, but this sentence seems to generalize it a bit too much. As a developer, I'm bottlenecked by ideas, not by coding. And I don't mean "make a site kinda like eBay" ideas, but ideas like:

"How do I optimally distribute quality across a video frame to best approximate the human visual system?"

"How do I measure video quality in a manner that takes into account the human brain's dislike for certain types of artifacts?"

"How do I reorder the operations in this 2D frequency transform to avoid having to transpose the data?"

The answers to these are certainly "ideas"--but they're ideas that are far harder to come up with than their implementations. While "make a site like eBay" takes a few minutes to come up with and a few months of development to implement, something like the above might take an hour or even as little as 15 minutes to implement, but the ideas are few and far between.

Here's the anecdote for the third of the above:

Developer1: I found a way to make this frequency transform a whole lot faster, but I'm not allowed to say yet because my university won't let me.

Developer2: Wait, that can be done faster? I didn't think it could be done much faster. If it can, it must mean there's a faster way to...

30 minutes pass

Developer2: Here's a patch, it makes the transform 60% faster.

No speed advancement had been made in that extremely important and often-called function for over 6 months--and it took only about 30 minutes to make it that much better merely from knowledge of the fact that it could be done better.

Boyer-Moore string searching is a good example of a simple but non-obvious "idea" (search backwards; it's much faster).

http://www.cs.utexas.edu/users/moore/best-ideas/string-searc...

Boyer-Moore is far from non-obvious. I remember having the same basic idea when I was a kid, when I was programming on the C64 and before I had access to the internet or decent programming books. The details - e.g. jump distances based on whether and where the characters occur in the search string etc. - come up in trying to create a working implementation.

I tend to think that ideas are a dime a dozen, and actually doing it is what makes the difference - Boyer-Moore is one of the few counter-examples. But if you also had the idea... could it be that you're unusually smart?

The details of Boyer-Moore seem obvious to me, in that you bump into them in trying to get it working, even if you don't foresee them. They are like workshop improvements. Did you manage to implement a version of it, as a kid? Or is this another instance of showing that actually doing it is what counts?

Our "idea" legal structure (patents, copyrights, etc) is not about ideas at all. Its about commerce. All that matters is implementation in a manner that increases money exchange. Everything is optimized around that, or is supposed to be.

An idea itself is of little value. If it were, the really hard stuff, like math itself, would be patentable.

If you had implemented a simple string search, and then someone told you that it was usually possible to do a string search without examining all the characters of the string, you'd probably come up with some Boyer-Moore variant. Nevertheless it took decades for people to come up with it. It's one of those cases where the basic idea is the really hard part.

Binary search, among the very basic algorithms, is toward the other extreme. An incorrect version of the algorithm was published something like 20 years before a bug-free version was published.

Dragon's Egg (Robert Forward) is an excellent hard science fiction novel, in which merely from knowledge of the fact that it could be done better figures prominently.

Reply on news.ycombinator.com