How is it that 'Hacker News' still doesn't have search? | Hacker News Reader

How is it that 'Hacker News' still doesn't have search?

29 pointsReadmore47 comments

Come on guys, this is getting stupid. This site has been up for over a year, it's popular, there are tons of new stories everyday, and we still can't search! What's going on?

View on HN

What's going on is that I have a few other things to do. Like writing essays and dealing with 58 (so far) startups we've invested in. Since the main advantage of implementing search would be to save people from going all the way to Google (http://www.google.com/search?q=site%3Aycombinator.com+Readmo...), it's not a huge priority.

You can easily put a search box executing in-site search in Google. This saves the users the context switch of going to Google.com.

Google never spiders everything. I'd be really surprised if google had caught every single story that had been posted here.

I know I have had trouble finding things I've seen on news.yc when using Google.

I had problems finding stories from before the switch to putting all comments and news stories on item?id=x, because a search would go to a broken link. But the object numbers stayed the same, so if you find a story with a broken link, you can just switch comment?id=x to item?id=x.

"... I had problems finding stories from before the switch to putting all comments and news stories on item?id=x, because a search would go to a broken link. ..."

Deleted node probably. Means it's been deleted or removed by user and/or was probably edited because it was "spammish" or poor quality in nature.

"... so if you find a story with a broken link, you can just switch comment?id=x to item?id=x. ..."

This one I noted after some changes in the code probably about 2 months after I started using it. I save all my posts and the links started breaking (link checking). The term was changed from 'comment' to 'item' (guessing) because it was shorter.

Before the update, items were broken down into 'comment' and 'link' (not sure if link was the object, it might have been news) but their ID numbers didn't overlap so I assume PG combined them because their was no reason not to, except for breaking old links.

The links I was trying to find weren't actually deleted, I just wasn't finding them because I only recently figured out I need to change the link to 'item?id='

"... tems were broken down into 'comment' and 'link' (not sure if link was the object, it might have been news) but their ID numbers didn't overlap so I assume PG combined them because their was no reason not to, except for breaking old links. ..."

Thanks for that explanation. Hadn't taken the time to work this out.

An example: I'm running a startup from Melbourne Australia and read an interesting comment from someone in the same boat, including details of a meetup type event, a week ago.

I didn't bookmark it because I didn't expect it would be hard to find. But now I can't find it at all.

The poster mentioned Melbourne was the 'artichoke end of the world'. Googling for 'artichoke' on news.yccombinator.com has no results. How frequently does Google spider most sites anyway?

Meetup: http://twitter.com/mtub

bootload's comment: http://news.ycombinator.com/item?id=52955

"... The poster mentioned Melbourne was the 'artichoke end of the world'. Googling for 'artichoke' on news.yccombinator.com has no results. How frequently does Google spider most sites anyway? ..."

Yeah the search here sucks. My attempt at finding stuff is to

A) save it on my blog so I have the original copy I can then search via google

B) in FOX: site: http://news.ycombinator.com search+term

Look one way to get around it is for someone to spider the site and link back (damn I lost the link .... someone has a search site where???), another is to get a google box for what USD$5K and integrate google search. As the site grows, search is becoming important for historical articles, phrases or people who you want to ascribe a comment. Guess this is what's meant by users loving new features. I'm just glad there is somewhere for stuff like this to be discussed. So while search sucks, it's not integral, for me at least.

And by the way the word "Artichoke" is a vegetable reference from Wallace & Grommit, "The Curse of the Were-Rabbit" ~ http://en.wikipedia.org/wiki/Wallace_%26_Gromit:_The_Curse_o...

Seconded.

Why not open source and let hackers on here help ;)

Beautiful comment!

brand dilution.

Someone has to make sure the hackers aren't hacking things up.

(Btw, how about a couple synonyms for 'hacker?')

I don't doubt that you have many other things on your plate but surely there is someone at YC that could do this. Or if nothing else take axod's comment and open it up so that someone on here can do it for you.

Update - Make it part of the Winter YC application. Post the data we need to know and let people write the code for you.

"Make it part of the Winter YC application. Post the data we need to know and let people write the code for you."

Are you just saying that because you want to see the Arc specifications? :P

Hahah... no but that's a good idea. Shhhhhhush! ;)

"Make it part of the Winter YC application."

As a way to weed folks out, I guess? Automatic disqualification due to inability to focus on ones own problems?

"... but surely there is someone at YC that could do this ..."

the headcount is pretty low, who? My idea would be to open up the data, release it as RSS, ATOM or XML. Then you could search it yourself.

"... By adding rel="nofollow" to a hyperlink, a page indicates that the destination of that hyperlink SHOULD NOT be afforded any additional weight or ranking by user agents which perform link analysis upon web pages ..." ~ http://microformats.org/wiki/rel-nofollow#Specification_2005...

Damn I just thought of one problem though. "Rel=nofollow" on low priority submissions means google will have low(er) priority search? Is that right?

Thank you for confirming my belief that putting up with the JVM or the CLR is worth it because you get awesome libraries like Lucene, Nutch, and most other things you want but don't want to write yourself or come up with an unsatisfactory workaround for.

Yes there's Google, but Google lacks transparency and if something doesn't index they won't tell you and there's nothing you can do about it.

Your best bet for searching Hacker News for now is http://nycs.bigheadlabs.com/ (an external util)

Thanks. I knew I had seen a post about an external search tool but since I couldn't search the site I couldn't find it again. ;)

ycsearch.com is not as fancy, but it is a bit easier to remember. Although, I guess it is just as easy to remember site:news.ycombinator.com, which it is functionally equivalent to.

but it doesn't pass the "artichoke" test ~ http://nycs.bigheadlabs.com/search1/?q=artichoke

http://www.google.com/search?q=site%3Anews.ycombinator.com

You're a hacker right? You should be building a search tool yourself! :-)

Kidding, but yes, I share your frustration.

I wonder how easy it would be to make a firefox plugin that let you easily search the site of the tab you're in...

Now that would be awesome. Even better if you could configure your own search engine. I use scroogle, but site:currentsite.com still works...

A quick hunt[1] didn't throw up anything for me... [1] https://addons.mozilla.org/en-US/firefox/search?q=current+si...

google toolbar has a "search this site" button

i don't want a toolbar. The firefox search area is about all i'm willing to use. It would be nice if pressing ctrl-enter to submit (or something as simple) would just search the site.

It would also be nice to use another key like alt or shift to dictate whether to open a new tab or use the current one.

My guess is that this takes an extension, not a plugin.

I have one made, just trying to figure out how to share it.

I can't find anywhere on the Mozilla development site that just lets you upload the xml file. They want to make me fill out a form that builds the plugin for me. Problem is that their form isn't really powerful enough to actually make a decent plugin. What am I missing?

For now it has been uploaded to: http://www.lorebroker.com/ycombinator.xml

Just download and stick in your searchplugins directory (C:\Program Files\Mozilla Firefox\searchplugins in Windows) along with the other plugins you find there.

I have created a plugin, that can be used for searching ycombinator.It is available here: http://trk7.com/yc, on just needs to click on it to install in FF and IE7.

I went ahead and created a Google Coop search here:

http://www.google.com/coop/cse?cx=013284775634912373901%3Avm...

Google has already indexed this page. Pretty impressive. One quick hack would be to put up a search box that redirects to a google search.

http://www.google.com/search?hl=en&q=site%3Aycombinator....

Google seems to like news.yc. I week or two ago I did a search for something obscure without using site:news.yc, and it returned a post from this site that had been posted something like 48 minutes before my search.

I remember a few instances of people submitting google searches as stories and within an hour the news.yc story was showing up in the results.

The recent stuff is easy, but google seems to not do as well once pages aren't already accessible from the front page.

With so many hackers here, why doesn't someone familiar with Lisp write a simple search module and send it to PG who can just check it for sanity and plug it in? I know its not easy to do without seeing PG's source code.

PG how about making news.yc open-source so we can all work on it?

Use Google.

Search is hard to do if you hack all the code (no external search engine or a database).

I could be wrong, but I don't think the site has been up quite a year yet...

Yeah you might be right, it sure seems like it's been that long. I'm sure it was here in 06, 9 months is plenty of time to put in "select * where title like '%query%';"

The site launched in Feb 07.

Doing search right is not just a matter of searching titles. And unless you're prepared to work hard to get it right, you're probably not going to do better than Google.

Adding a sitemap might help Google index the old pages better. Adding date-based archive links (like many blogs do) would probably help as well (making the site structure more like a tree and less like a list).

https://www.google.com/webmasters/tools/docs/en/protocol.htm...

I agree, a Search will be quite helpful

Reply on news.ycombinator.com