P16: a blog by Matt Kangas home archive
06 Feb 2008

Erlang documentation poorly SEO'd?

In the brief time that I've been watching the Erlang world, I've noticed several cases of people saying:

The only example I have handy is Russell King's post on Jan 22, where he says, "For whatever reason, google search didn't give me anything."

In my experience, when a search on an obviously-distinctive term fails to yield an obviously-canonical website, the problem is usually that the website is poorly search-engine optimized.

Years ago, I worked at HotJobs.com. When I started, if you went to Google and typed "hotjobs", the results showed our homepage, our discussion boards... and nothing else. Why? Because the most valuable part of our website – the job listings – was not accessible by clicking links. Google couldn't crawl our most-valuable content, and thus it didn't appear in Google.

We eventually, collectively, came to our senses and fixed this bug. Shortly thereafter, we started getting a lot more traffic courtesy of Google, and everyone wondered why we hadn't done this sooner.

I suspect that something similar may be true for the Erlang HTML documentation. It's not hidden behind a search form, but it may be poorly browsable by a web crawler: frames are involved, links seem deeply-nested, possibly lacking helpful keywords. Long documents also tend to be penalized by search engines. Maybe something else is amiss.

Consider for a moment the PHP documentation, which has one page per function call. Choppy, but quickly browsable, with short and descriptive URLs. Search for "print_r" on Google and you get the PHP manual as result #1.

Searching for a similarly-obvious keyword in Erlang yields strange results. "io:format" returns www.erlang.org/doc/man/io.html as the top result, but "erlang io:format" drops it to 2nd place. Wierd!

That's as far as I've investigated. Google's Webmaster Guidelines are a good starting point for further investigation...