HN Explorer

Ask HN: What would you like to search and can't?

23 points by diego ↗ HN
Here's something that keeps me up at night. My company (IndexTank) makes it possible for people to build custom search engines. I've been hacking a few pet apps with our product to search stuff that's important to me (e.g. our irc channel logs).

There are many apps out there with poor or no search (e.g. HN), but in most cases you can make do with Google (searchyc in this particular case).

What data/apps are out there for which there are no acceptable search options?

28 comments

[ 3.1 ms ] story [ 73.8 ms ] thread
bartonfink
The dark depths of my soul.
diego
No problem. Do you have a REST api? :)
Blankwood
Wha about search for sites like posterous or Tumblr?
diego
Posterous is an interesting one. I'd build a whole-site search function if they had a firehose-like api call, but as far as I know it's only possible to access content for an individual blog.
atgm
One of the major problems I have is that I talk to my friends across all kinds of different media: Facebook, Twitter, AIM, MSN, e-mail, forums... sometimes one of us says or links something and I'd like to be able to search everything, but I have to search each medium separately.

Privacy concerns probably make this kind of convenience impossible though.

diego
I've been thinking about that a lot. Greplin is supposedly working on that problem but it's impossible to sign up these days.
frou_dh
Web page source code (not just visible content).
phlux
What would be interesting would be an HTML !diff type search: Find all the pages that are coded similar to this [example].

This would allow comparison of code and see how others do things.

cousin_it
Not sure if this reply will help you, but I'd like to be able to "search" the physical stuff I own, especially the documents. Could it be possible with some sort of RFID tags?
diego
The search part would be the easiest. The hard part is keeping track of spatial coordinates for all those items in near real-time. If you could tag say 1000 items in your house and gave me that api, I could set up a web app for you to find them.

Does anyone know if such technology is readily available/cheap?

blago
Full text search (HTML aware) on my browsing history. I'm working on it :-)
santip
Google Chrome already features this in the Bookmarks > Show Full History view. Make sure to close the page before searching for it otherwise it won't be indexed yet.

Perhaps you mean something else by "HTML aware" and this is not what you're looking for.

blago
"Google Chrome already features this"

This works great for searching your Chrome history on the current computer. Throughout my day I use multiple browsers on at least 3 machines.

"HTML aware" is exactly what I'm looking for. I want to be able to limit my search only to links or headers or maybe even text within elements with class="content"

wrath
It's not really what I would like to search, but really how I would like to search. Adding filters or restrictions is becoming very important to me. For example, I want to search google and I don't want my search results to return any stackoverflow copy cats, or I'd like to be able to group my results by something (e.g. domain, relevance, date, etc). Also, I'd like to specify what to search (e.g. search title only, h1 only, body only, etc..)
blago
I agree. There are no good search options for power users that want to get very specific.
dy
All my reading history across various channels. I think I read news in Hacker News, Instapaper, twitter and Google Reader. I have on many occasions said "I just read this cool article, let me find it for you..." only to spend 20-30 minutes trying to find it with Google site: incantations.
ericb
Source code and programming related searches are often problematic. For example, "-" is a search operator, and so is the word "and". Special characters are dropped by default. Also, specifying the language results must be in would be handy. There has been some work on this (koders.com) and you can work around it on google with quotes, but I'd say there's no killer appp yet. Some query examples:

foo -bar (where bar is a command line option)

foo and bar

@username

session[:user]

"rails nested routes"

One particular problem is that google has an anti-youth bias, so very often, I get old doc or results for an error or tutorial unless I'm careful. For example, "rails nested routes" returns old rails 2 route info unless I add in "rails 3"

[dead] k0nad
Also google has "search by date" in advanced
secret
Math related search is generally not good. You can find theorems and definitions but proofs (non-famous ones) exercises are difficult to find.
happybuy
All of my phone/voice conversations that I have ever conducted through my iPhone.
NY_Entrepreneur
About half of all the content on the Internet.

What content is that? The part where keywords are from poor down to useless to characterize what the user wants and means in their search.