If anybody wants an easy way to set up their own HN, may I suggest taking a look at Telescope?

http://telesc.pe/

It's a real-time, open source HN clone built on Meteor, with features like invite-only mode, notifications, and a lot more stuff.

And since it's open source, if something is missing you can always code it yourself and contribute it back to the project.

This is pretty cool! I've been working with Meteor for a few weeks, but haven't heard of it before.

Forking!

How do you install meteorite? I can't get this working!

Frustrated!

Thanks. I got this far after posting my comment. Figured out I needed NodePackageManager. I hate when instructions say npm and dont say what npm means. Now I am getting another error when I run mrt inside the telescope folder. I am working on it.. I love this. I spent weeks trying to get newscloud going, but never did...

Text is dispersed too much. I prefer HN because you can read a lot more in a packed space.

CSS is easy.

Does this have karma style points for users?

Not yet, but it's coming.

I've done similar.

Bayesian filtering seems to work great on HN headlines. I trained mine with about two years of data scrapped from http://www.daemonology.net/hn-daily/ (apologies to whoever runs that), then just basically wrapped that with some code that grabs hacker news's main page and displays the filtered headlines to me. It nails politics and startup crap with stunning accuracy.

The only problem is that now I find myself using both that system and the website itself.

Anywhere I can get that training set without DDOS'ing some poor guys mirror?

And if you build the next HN, please add an API

Technically, HN already exposes an API over HTTP. It just uses HTML as the exchange format instead of XML or JSON.

I try to remember that line the next time my boss wants me to code an API.

that's a shitty excuse. HTML as a "protocol" mixes presentation with content. An API should not only be presentation agnostic, but presentation free.

Where then do microformats fit, in your opinion? A defined standard for API-like transmissions, but mixed into presentable content.

Hmmm, not sure if I agree with that or not... but if so, it has a very limited usage quota before you get banned. :)

What API do you want? Automated submissions of posts is not a good idea, so that leaves data scraping.

If you're looking to extract data from HN, use HNSearch. http://www.hnsearch.com/

There's no reason to scrape data from HN itself.

HNSearch is great and I use it for my Wayback Letter project, but it hasn't been around that long so I couldn't use it before. When I started Hacker Newsletter about 2.5 years ago there was nothing, so scraping was the only solution. One of these days I will convert the application I built to build out each issue, but there also is a risk of HNSearch going away just like the last search engine did.

I'm curious why you disagree, aside from the quota thing? I have written many API consumers over the years and there is basically no difference between parsing HTML or any other format; XML especially, for obvious reasons.

HTML APIs to come with the added benefit of being less prone to change, which I realize goes against conventional wisdom, but seems to hold in practice.

I agree that parsing can be easily done, although I don't think it is necessarily equal... but I guess that would depend on a case by case basis since a lot of API's are terrible. What makes it not an API IMHO is that you can't consume it when needed, but rather you have to consume it all the time. I guess you will call that a streaming API though. :)

HTML APIs to come with the added benefit of being less prone to change

Really? I don't think that does hold in a lot of cases. Using HN as an example, it broke a year or so ago when PG changed how the job postings were listed. Again, the quality of an API can vary, but at least you would know what changed in that case.

Also, make sure that every time someone performs an action on the site, you generate a random number. Add that random number to the number of seconds since their last performed action. If that number exceeds a threshold (that is assigned randomly and valid for a randomly assigned amount of time) delete a random part of the session.

What are the main reasons you want an API?

Building more powerful clients for desktop and mobile. Think user tagging, comment filtering, custom sort order, interface improvements.

I really like that HN still has such an old, simple interface. It's one of the reasons I haven't left yet.

About a decade ago I noticed this phenomena.

Worthwhile communities need to be maintained by barriers. The kind of barrier that I personally enjoy most is, "Crappy enough interface that only people who really like the content will be motivated to show up."

I think it just correlates with the presence if images. Also StackExchanges have excellent UIs, APIs and still very good content. StackOverflow itself could also be worse.

That is a good comparison. StackOverflow has a good API, but maintains itself with carefully thought through and zealously enforced moderation rules.

I grant the existence of useful content there and a good API. However after one too many brushes with how differently moderators think about stuff than the people who I liked talking to did, I have demonstrated the truth of my comment by choosing to leave.

StackOverflow works for a lot of people. But for me, personally, a stripped down interface and less moderation is a better experience.

Yep. No magic tricks. Just substance.

Amen to that.

Or just create one themselves: http://harpb.com/static/news/news.html#/featured/p1/100

Presentation, along with content, is a better experience than just having good content.

Never really thought of it like that, a statement worth exploring for sure

> I really like that HN still has such an old, simple interface.

You having the interface you want shouldn't impact my having the interface I want.

An API allows everyone to have the interface they want.

I guess something like the possibility to have a mobile application (either web based or native app) which would be usable (the tiny up- and down-vote arrows of HN are not easy to use using a finger on a touchscreen) and which would actually work by not relying on html-parsing or search kludge.

The best reading experience I've seen (by far in my opinion) for iPhone/iPad is http://cheeaun.github.com/hnmobile/

If you have an iPhone, there's news:yc. Great reader. Not perfect, but the best I found so far.

>> I guess something like the possibility to have a mobile application

See http://ihackernews.com/

That's a website not a mobile application.

>> mobile application (either web based or native app)

Sounds interesting, can you give us more details on how it works? I had a quick look at your homepage, but couldn't see any links for it, or a github profile.

I've thought about designing a community with script hacking in mind (a simple initial layout with an api) and then letting users be able to script it themselves through their accounts. If it could even be done safely and sanely, everyone could have the features they wanted by writing their own.

In lieu of that of course there's always greasemonkey.

Sounds interesting. Would love to check it out.

Would you be willing to share this tool?

Yes, I agree, build it.

However, please do not make it like HN or you will fail.

Be different. Brand it. Have a personality.