Is it possible to change the 4-letter search limit?

[Stoid]

WAY too many meaningful words are three letters! Two most recent searches that failed because they were 3 letters: CSS and FOX.

Wah.

[TubaDiva]

Searches are about the most intensive thing the system does -- we've always had to balance the resources used for searches versus all the other needs of the site, including rendering pages and writing posts.

We settled on the limit that we did because that seems to be the best balance of needs and makes the most equitable use of resources. If we went down any further it would affect everything else the system does in a profoundly negative direction.

You might try using Google for your more intensive searches and bypassing the faulty SDMB search engine altogether. Please see this excellent posting on the subject here:

http://boards.straightdope.com/sdmb/...4#post11650034

[Peter Morris]

Has Google indexed the whole site yet? Last I heard, a Google search typically missed about half the posts compared to a similar board search.

[Vinyl Turnip]

It hasn't bothered me yet on this site, but this same issue has caused great me frustration when trying to search other fora, particularly technical/product support boards. In some cases it renders the search function practically useless--- making it impossible to search for (most) file extension types and a great number of acronyms.

[Giles]

I gather that the problem is that the limitation has to be on number of letters, and not on a stop-word list. If you could search on three-letter words, you would require the system to search on very common words like "the" and "and", which are not terribly useful in searching. I suspect that the use of stop words would require major work on the software.

[TubaDiva]

vB says:

Search Index Minimum Word Length

When using the vBulletin default search, this option limits the size of indexed words. The smaller this number is, the larger your search index, and conversely your database is going to be.

Including three letter words makes the database insanely large.

[Munch]

Originally Posted by Peter Morris

Has Google indexed the whole site yet? Last I heard, a Google search typically missed about half the posts compared to a similar board search.

Where did you hear that? I can't imagine that being the case after about an hour of Google starting to index. A thread with 10 pages with the word "dominoes" on each one is going to come up with one hit on the SDMB, 10 hits on Google. In fact, googling "site:boards.straightdope.com dominoes" has 5390 hits, searching "dominoes" on the board comes up with 471 (bad example, since there's a thread in the Game Room with thousands of posts with "dominoes" in the title). "Congresswoman" gets 282 on the board, 2620 on Google.

[GuanoLad]

I thought there was an "exceptions" list in vBulletin, where you can add in some relevant three letter words it can search. I used to look after a board where some three letter acronyms were common (DVD for example), so included many of them as searchable.

[MsWhatsit]

Google seems to have indexed the SDMB pretty thoroughly at this point. I can't think of specific examples, but there have been a few times lately that I had a vague memory of a really old thread and did a Google search based on a few keywords I remembered from the thread, and Google had it.

[TubaDiva]

Yep, there is. "Words to be Included Despite Character Limit" What's your pleasure?

[Guinastasia]

Actually most boards I visit have a four letter search limit. So it's not just the dope.

[Stoid]

Originally Posted by tubadiva

yep, there is. "words to be included despite character limit" what's your pleasure?

dvd
css
fox
nbc
cbs
abc
msn
cdc
cms
vip
mp3
cgi
ram
cpu
php
sql
lcd
jpg
gif

[MsWhatsit]

Mac.

[TubaDiva]

Okay, I've added these in. Let's see what happens.

[hajario]

Very cool. Thanks for adding those.

Did you know that Stoid has only used "Fox" in a post three times and not since 2003? I would have never guessed that.

[johnpost]

companies that might be searched for might include

ibm
3m

dozens of past and present government agencies (in the USA)

and then there are lots country abbreviations that have two or three letters, for example.

usa
uk
prc
uae

[Chronos]

A few more:

OSX
PS2
PS3
Wii

[johnpost]

also have been messages on

dos
os2
cbm
c64

and chemicals like for example

co
co2

and standards like

nec

[Chronos]

Really, I think it'd be easier to allow 3-letter words by default, and just expand the badwords list. Most of the problem with searching 3-letter words comes from a very small number of repeat offenders like "the", "and", and "for".

[TubaDiva]

Chronos, that's not how it works. You have a point but the code writers don't always go along what seems to be the logical choice. YMMV, I guess.

I've added the latest suggestions. Thanks to everyone for their help.