I'm not actually sure if this is possible, but could you make it so the google search always adds "intitle:" to the search so it works a little more like the old search?

Especially so it doesn't catch, say, the topic to rename a trope over the trope itself.

Also, could it be tweaked so, that it would more often get the main article, instead of the discussion page.

You can't, to my knowledge, tweak individual Google search results in any way. The only major option is to either use robots.txt (with some non-standard Googlebot patterns) to disallow bits like ?action=history, or to use the <meta> tag for robots (likely using noindex, follow) on pages you don't want indexed.

Closest you can get to eliminating the forum discussions (without blanket disallow statements, not letting them ever be indexed) is using inurl:pmwiki.php - but that seems to reorder the results a bit. Can also do fancy bits like 'intitle:"The Search Phrase" inurl:pmwiki.php' - this seems to be the best option currently to find a specific trope and nothing but a trope.

Fast Eddie already did something to cut history results and such, but it takes a while to come into effect.

I'm failing to see any evidence of anything done to the history pages. There is no meta robots tag on history pages, nor does the robots.txt file contain any disallows.

Well, he mentioned it, that's all. I could be wrong.

Oh, and couldn't forum posts be removed by disallowing posts.php?

edited 24th Aug '09 12:07:50 PM by Tzetze

I don't know what intitle: is, but it sounds like a bad idea. The advantage of the Google search - and the reason we got it in the first place, IIRC - was that sometimes the most obvious search terms aren't in the titles.

Tzetze: Yes (well, with the correct paths - since there's several files involved), but that will entirely eliminate the forum from Google's index. It depends on if that's what's desired or not. I can see having a Google text search of forum posts being a good thing.

Arguably the best solution, if you want to segregate search results to that degree, is subdomains (i.e. forum.tvtropes.org). I believe Google custom searches can easily be limited to a certain subdomain, but this would be somewhat more difficult to manage than the current setup, and would require a fair bit of rewriting of the current forum system to 'detach' it from the wiki more.

Bobby G: intitle takes the next word (or phrase if in quotations) and searches page titles only, for that text. You can combine it with other search modifiers, such as doing 'intitle:Xanatos index' would find all pages with the title "Xanatos" and then search the text of those pages for "index". Could also do 'intitle:"Off The Rails" troper tales' to find pages with "Off The Rails" (as a phrase) in the title, and then "troper tales" anywhere in the page.

Bad idea to use it for a 'default', but good trick to use for more complex searches.

Perhaps not automatically SafeSearch the thing.

I can't search for Magnificent Bastard without getting a bunch of wrong answers. I can get around it easy enough, but I'm so used to the old search that I always have to look at that page with The Magnificent Seven Samurai on the search results first.

edited 24th Aug '09 6:31:11 PM by Zephid

Oh right, that too.

It's impossible to find Fetish Fuel (and a bunch of other tropes) with the google search because it's always SafeSearched. Really annoying sometimes.

Append &safe=off to the URL after a search.

Alternately, I've done a bit of local rewriting to the document (go fancy-ish HTTP proxy) and dropped the following into the HTML for every page I view, in the correct form for the search:

<input type="hidden" name="safe" value="off" />

Should someone with the appropriate template access be inclined, this is a site-wide method to disable safe search. Given that there aren't pornographic pictures, and there are very few attempts to be safe for work, this shouldn't pose any real sort of problem.

More simply, would it be possible to have the Google search bar on the watchlist page?

That's usually my first reflex when I want to search a title on TV Tropes while I don't have any tab open...

It seems like when searching, all the proper articles show up as "The 'Main Foo' trope as used in popular culture, with a list of examples from all media." Is there any way to stop that?

I realize I'm bumping an old thread here, but I've been wondering — why exactly did we start using the Google custom search on the main page instead of the title search?

I realize it's been that way for a while, and I realize title search is still there under tools, but it seems very visitor-unfriendly to have it out of the way when there's a great big glaring search box at the top of every page which isn't as effective.

Is it a technical thing?

Server resources most likely. Google is external and doesn't put any strain on the server, as the old search likely tended to. To get an efficient, internal search system setup is a lot of work, and often requires a specific set of tools and knowledge that aren't always readily available.

Google is also, in my experience, superior if you know the tricks of how to use it. It's not as nice for simple title searching, but it's a lot more powerful if you don't know the title of the trope. (And if you're looking for a specific trope, you can try searching for 'intitle:"Trope Name"' — it usually works very well like that.)

Google search was originally embedded in order to make the wiki just that little bit easier to navigate. The wiki's own search was taken off the page shortly afterwards in response to DDoS attacks.

^^Not long after the Google search was added.we started being hit by search-flood D Do S attacks; full-text search requests numbering in the 10s of thousands per second. That regularly took down the servers, and the only way our techs could stop them was to take the site search offline. Eventually, the problem was turned over to a security consultant, who did some sort of magic that got the site search back, but they alsoadvised that it be moved to somewhere other than the top of every page. Why, I don't know — I'm not tech-y enough. But the end result was, the Google search is at the top of every page, and the site search is in the Tools menu on the sidebar.

You can see the queries per second here.

Ah, thank you, all. That explains a lot.

I don't really mind using the Google search bar myself, it just makes me a little sad in terms of the loss of first-time-user friendliness, since the title search seems like the simplest way for people to get at our content for purposes of exploring.

But what must be must be.

@Lmm: See (and bump) this page.

Edit: Hey, the old search URL now correctly redirects to Google instead of the "search is disabled" page! I would say "at last", but It has been quite some time since I last tried.

edited 15th Oct '09 1:40:44 AM by Medinoc

I hope something has been done about getting history results in the Google search. It's getting annoying.

But thanks for letting me know that the regular search still exists. I assumed the replacement meant it was down. If the hackers think like me, then we'll be golden. If not, we can simply add some sort of delay in searches, like other places do.

Adding "-inurl:action=diff" to the search prevents history pages from appearing.

We started telling Google not to index the diff pages a month or so ago. The diffs that are in there are from before that change. It will take a while before they are purged.