Members should be able to search short words

AIUI, the reason for limiting searches to words greater than three characters was to prevent people tying up the server (maliciously or innocently) by performing repetitive searches on small and/or common words. But Members pay to be here. We have no reason to bog down the system. But we may want to look for subjects that are made up of short words.

This restriction should be lifted.

I suspect it’s more the indexing of small words rather then the actual searches. If you are willing to do that, opening it up to the masses shouldn’t be a big deal. But the indexing might be the killer.

I think the actual searching is just one part of that. I believe indexing 3-letter words would greatly increase both the resources needed to do that extra indexing and the size of the index, and that would kill performance.

vB already has a file of common words not to be indexed. Some are short, like “the”, and some longer, like “should”. I would have thought that that would be sufficient. But the pros say otherwise, and I am not one, so I yield entirely to them.

Given that any time they allow searching at any significant level these days, the server goes quite wonky, I’m just grateful that the search feature hasn’t been turned off. Don’t laugh, it happened in the ParadoxPlaza forum (makers of the Europa Universalis and Diplomacy games), for example. :eek:

Sure, but you could limit the impact of that by not including the 17,576 most common three letter words in the index.

I would be interested in seeing a list of longer common words. I can’t do a vanity search, because “itself” is not allowed.

So stick quotes around “fear itself” in the search box.
That’ll get you things like this.

But if I put quotes around, say “top gun” I get this:

I suggest you change your username to Toast Museum.

Why would three-letter words be any greater burden than any other size, though? I mean, sure, “the” and “and” and the like are way too common, but that’s what the badwords file is for. What is it about (say) “OSX” that makes it a worse word to index than “museum”?

Informed speculation based on behaviour of my board, which is older: there are a large number of occurrences of the 3-letter words, and this makes the index large. And searching through badwords.php every time is ponderous (relatively speaking), and having a large number of exclusions slows things down. There is a cost-benefit which they weigh, and it turns out that in the opinion of the folks who made the default setting, 3-letter words go over that cost-benefit boundary. Given the incredibly slow performance of the current server/setup, and the fact that the Search doesn’t work all that well anyhow because it needs re-indexing (or repair…), I’d guess that changes to Search are a non-starter.

What she said.

The finer the filter, the greater the burden on the server.

From vB: “The smaller this number is, the larger your search index, and conversely your database is going to be.”