Hacker News Reader

Comment by bravura | Hacker News Reader

The most important problem to solve is that when you do a search, or browse by a tag, that no one slips through the cracks. With a friend organizer, the worst possible thing it can do is become central to organizing your social life, but then silently "forget" about certain friends of yours.

I've thought about this very application (I call it "CRM for hypersocial people") and the most important component is the natural language processing. In particular, you want good autotagging and query expansion, with an emphasis on recall.

More specifically, if one of your ballet dancer friends is manually tagged "ballet" but not "dance", and you search for "dance", you want this friend to appear in the results.

Additionally, if you are looking for "hackers", it should know that it should include people who are "hardcore programmers" or maybe those that program "ruby".

View on HN

The question really is: What problem are we trying to solve?

The answer is: Allowing you to easily find certain types of people over your hundreds or thousands of contacts. Types = certain business or personal relationship, certain skill, certain place you met, etc.

Here are the core features, the minimum viable product. In fact, I would avoid something that is bloated and doesn't focus on these core features:

* Automatically importing and deduping your contacts from a variety of sources: email contacts, linkedin, facebook, maybe twitter.

* Automatically assigning a rank score to each friend, based upon your frequency of contact with them, and analysis of the social graph. You want your spouse to be ranked high, and the guy you met once at a networking event to be ranked low.

* Depending upon interactions with people, you can upvote or downvote them, to change their rank score. Ran into someone you want to hang out with more? Upvote. New girl made a date with you and then stood you up? Downvote.

* You can search or browse by tag. There is smooth autotagging and query expansion, to make sure searches don't miss certain friends.

That's it.

Why did this get downvoted twice? It's completely reasonable, and I happen to agree that it would be useful (and not just for 'hypersocial crm')

I originally offered to share the technology, either free for open source apps or as a business deal if you're using it to make money. I imagine people found that offer distasteful, so I removed that last part from my comment.

I personally included it because I think more widespread dissemination of NLP could improve a lot of technology. But I can understand why people would think that I made my comment merely to scurry up deals, and not to inform them.

I'm surprised that Fred just didn't ask the etacts guys (https://etacts.com/) for the features he's looking for (or maybe he did and they rejected the requests).

I tried to implement the same thing a few years back, but I got lost in feature creep. Maybe I'll do a quick relaunch to address where I went wrong.

i agree with your first point. when it's time to invite people to the christmas party, i don't want to leave anyone out accidentally

Reply on news.ycombinator.com