Interesting... it seems you must have some other criteria because my (admittedly woefully outdated) profile lists my (admittedly neglected) Twitter account, but I don't see myself on there. ~5000 karma should put me only a couple pages in.
Note that the inference algorithm is quite forgiving, so "twitter @Locke1689", or "@Locke1689 on twitter", or even "twitter: Locke1689" would all have worked out.
About 50 people (from a pool of 1000) with twitter accounts on their profile were left out due to this not so optimal inference.
I'm not on there even though I have a direct link to my twitter account. If I was left out on purpose then that's fine, just letting you know in case it's a bug though.
I think you could have easily assumed any word starting with @ is a twitter profile, especially if it does in fact correspond to an existing twitter profile.
Not being very thorough also allows people to self-exlude from the listing. Which seems be in the interest of some.
That's a great point. Can you post info on the site on how to be included or self exclude for those who have a twitter account? (I don't but have considered getting one. I would kind of like to know how to self exclude should I get one. I've had lots of issues over the years with unwanted, negative attention. One of my primary concerns these days is having some control over how much exposure I am getting.)
It is schedule every few hours. The biggest issue now is twitter limit request, which is what I'm solving right now.
Also the attention is not that bad (8775 visits so far). Also, heroku's varnish cache on static assets, html page and on Ajax requests is pretty great.
And mixpanel is handling the metrics, so I'm pretty covered.
No its because he's using HN search. I would venture to guess that most of your karma is driven by comments, for which HNS records as 1 karma (comment karma is hidden, so HNS can't get that info and assumes 1 )
This might argue for making comment karma viewable after a thread has existed for, say, a month: at that point, the absolute scores probably don't matter much for fresh readers, but the raw data could still be useful for efforts like this one.
