Posterous (YC Summer 08) launches search for every Posterous blog | Hacker News Reader

Posterous (YC Summer 08) launches search for every Posterous blog

blog.posterous.com

23 pointsrantfoil30 comments

So are you guys going to launch every feature on Hacker News? Think about it when every startup starts launching their each feature announcement on Hacker News?

Just because you are a YC alumni startup, does it make sense to launch every feature here?

Another thing which annoys me about YC startups is - to get immediate attention, they add "(YC Summer/Winter 08/09)" in the title of their post, and other YC alumni vote up immediately. And the post comes to the front page. I'm tired of these tricks!

Please, there are many startups launching many features every single day. Launching search feature is not a news. Launching a startup is a news. So I would appreciate if you can limit such releases to your blog only. If we are interested, we will follow your blog for feature releases.

Please don't take this personally. I meant this to all YC alumni startups who launch features as if it's a big press release on Hacker News using all above mentioned tricks.

Anyway, it was my $0.02s. Hope you agree to it.

Startups making progress and launching new features is my second-favorite thing to read on news.yc. New startups launching is less interesting.

My favorite thing to read is user feedback on new features.

I think new startups launching is more interesting than new features, but I do think new features are great to hear about.

I actually enjoy hearing about feature releases from other YC startups. I accept your feedback but respectfully disagree. If you don't like it, please don't upvote it.

My small advice to Posterous team on search:

1- Work more on equimatch before doing the partial match. e.g.: if I type ‘ruby on rails’ you should worry about returning post with ‘ruby on rails’ first before partial matches. In your case the second and third results do not even have occurrences of ‘ruby on rails’.

2- Your match text algorithm, the one responsible of coloring match text on result in green should focus more on full match for now until you find a way to do partial match effectively. One hint, use a regular expression algorithm, that’s all the help I can give you for now.

Overall you guys are doing a great job.

edit- One more thing here, if I search for 'iness' you return 'business'. This tells me that your going the wrong way on search. You should do more research on the subject of search.

We used to do prefix and infix search, but the amount of value provided was not huge. It also really blows out your indexes / indexing time. I'd like to add this back at some point, but we're not there yet.

As for supporting equimatch using quotes -- its actually not too hard to do this. Just wanted to ship early and iterate rather than expend too much effort getting it perfect.

A plan violently executed today is always better than a perfect plan executed weeks in the future. But thanks for the feedback -- these are important search items that need to be addressed.

If I search for ‘iness’ and you return ‘Business’ I wonder what your index look like. If it is what I think, you are going the wrong way rantfoil. This will be very time consuming if you are doing it from scratch; it is worth looking at Sphinx as thorax suggested or any open source.

Oh, we do use Sphinx. It's excellent. It would be insane to write your own full text searching or rely on MySQL's built in indexing.

Or use something like Sphinx (or Solr, maybe) which does a lot of the work for you in building a decent search backend.

Has nobody told posterous yet that they need to provide themes for the pages? If I were them, that would be the number 1 priority, not search.

Themes are something we have to get right -- we've got about half a dozen themes already designed and we're starting to work on the feature in the coming weeks.

Though I venture to guess early facebook users may have asked for themes for parity with MySpace, and it didn't turn out to be such a big deal in the longer run.

Thanks for the feedback -- you're right that it remains an important feature.

Well, for my part, lack of visual customization is the sole reason I don't use Posterous for everything I blog. You reeled us in with cool tricks, but I'm not sure how I can ever make it my core blog if it can't improve visually.

It's also one of the key reasons I haven't yet recommended it to my friends who aren't early adopters. My less tech-savvy friends who use LiveJournal and Myspace see the service more as a geek toy. I showed my wife how easy it was to get content added and she loved it-- but she'd never tell anyone to actually go there unless she can put her personality onto the page.

For my part, I don't care so much for custom pre-built themes-- I just need a way to add my own CSS.

And then I see responses that don't agree that this is the #1 thing on Posterous's plate and it makes me sad as a user. I know how unfun that kind of customization is to code, so the geek in me can relate 100%-- yet speaking as a user, let's just get this feature already and let that be the one I see announced on HN.YC instead of these other cool features you guys want to add.

Keep up the good work, though. You guys make good stuff, no doubt.

Many thanks thorax -- this is the feedback we need to hear.

That's why I stopped using posterous. I'm not the type of guy who likes themes usually, but it just felt so bland after a while. Nobody commenting, no personalisation, there is nothing that says 'me' about the page.

With facebook, I've got my friends status messages all around, so it's a 'me' site. But posterous is like a clinic - it's just so neutral.

Want to be on the private beta for themes? =) (EDIT: we don't do private betas. But trust me, its coming. ;-) ) These are all concerns we're working super hard to address.

One tough thing to balance about product design is that we want the experience to not require all these extra steps. Simplicity in visual design is one way to mitigate the 'tyranny of choice' that many people face with complex software.

I think in the long term this problem can be identified as the "low bar high ceiling" problem. The most well designed software is immediately approachable, but provides a ramp for users to create real, powerful, immediately useful, personal experiences.

The ramp is what we seek to build. Thanks for your thoughts!

Have you thought about putting a search bar on the main page? Or do you think that will take away from the ease of posterous?

Also, wouldn't you want the explore search bar: http://posterous.com/explore/

to look like this one: http://posterous.com/explore/?search=garry

Finally, on the explore page I think it'd be cool to have a running sidebar of the most recently updated posterouees (sp?).

Dan, awesome ideas man.

For the homepage, one thing we obsess about is keeping the choices dead simple obvious. (see tyranny of choice problem) We want people to try posterous, and I'd be worried a search box would be pretty distracting.

But the rest makes a ton of sense.

I second the "most recently updated" list on the Explore page.

I think Posterous has some excellent features going. What it doesn't have is a community feel behind it. I get the feeling that every Posterous blog stands alone. Versus other blogging systems - Tumblr comes to mind - where community is a big part of the fun. That's one of the things that's made Posterous tough to consider as a main platform - that and theming.

What do you mean? You can't even comment on a Tumblr! There is no community there by design.

Are you a Tumblr user? Because the large part of the community occurs on the Dashboard, through reblogging and passing things along. You get some people who comment back-and-forth, some people who act as nodes for large amounts of content. I'd love to map it: you get a brilliant web built up.

Tumblr isn't about the individual blog community. It's about the Tumblr community. And I like, actually, that when you look at a blog it seems entirely personal, but that from the dashboard it seems like parts of a larger community.

Also: see http://tumblelog.marco.org/61329864, where you can see the actions of an item in the community. They've also beta-tested a "questions" module, which you can see here: http://www.davidslog.com/58675697/where-should-i-get-cupcake...

The Tumblr founders are very against comments, and I love that. Comments don't form a community. They add a facade of a community in all but the best cases. Community relies on the form of the medium, and commenting very rarely is the answer. With forums, you have threads; with Tumblr, you have likes, reblogs, and answers. It's a wonderful solution.

The concept of bubbling up conversation to the top level of one's own published blog is fascinating. I didn't quite understand reblogging before reading your explanation, and I think that's the main downside to the scheme, but there is certainly value in making the act of commenting ALSO an act of publishing on one's own blog.

Yeah. It's tough to understand, and sometimes I feel that it detracts from the blog entirely (extended back-and-forth conversations start to look amateur), but it's a very neat system that gives you a lot of freedom in how you choose to handle interaction.

I just tried posterous -- it seems wierd that you don't need to sign up to create a blog, but do to comment on one.

100% anonymous comments are an option that can be turned on by the blog owner.

When anonymous comments are enabled, we run those comments through akismet and some other spam detection logic, but for many people, they just don't want the hassle.

What was Posterous developed in?

Ruby on Rails + special sauce

what did you use for search?

custom or acts_as_xapian/ferret/solr ?

Curious cause we had a discussion of rails search at the last local rb meetup

Was using sphinx + ultrasphinx, but was talking to Evan Weaver the other day and he mentioned thinking_sphinx is the one to use now since Pat Allan is really intensely working on it, so we switched over to that in this release.

Note: For larger production websites served off multiple app servers, thinking_sphinx doesn't work well since it tries to do reindexing of changed items directly within the rails process. However, there is a great fork of the latest plugin that can use updated_at, so you can index your deltas on a cronjob, offline. Check it out if you're going to use thinking sphinx: http://blog.edhickey.com/2008/09/15/thinkingsphinx-rails-plu...

neat application

Reply on news.ycombinator.com