31–40 of 71 posts
Earlier quoted context omitted.
Oh wow. Thank you for that explanation. It's impressive how many of these profiles were made. This means the total number of real HN accounts is probably significantly less than 28k.
I'd count "real" profiles by requiring that they either post a link, have at least one non-dead comment, or upvote a non-dead comment.
It would be really nice to have dataset available for researchers with username, their posts and their comments. As an example, we could use the "SPEAR Algorithm"[1] to find expertise in specific domains per HN user. [1] http://www.michael-noll.com/projects/spear-algorithm/
I like the guy whose name is user-id (:
I'd be interested to see some analytics, such as how many users are there with 100+ karma because that's likely a better measure of the number of active users than the number of registered users (many of which are likely inactive). However, that doesn't account for lurkers. Or, the number of users that have logged in within the past week though this is probably something pg would have to give us because I can't think…
For comparison, pg posted figures on Feb 9 showing 90,000 unique visitors (timeframe unclear to me, but I'm guessing daily) - http://ycombinator.com/images/hntraffic-9feb11.png If (as the results of my Google search revealed) there are 28,600 members, then 90,000 uniques is roughly a 2:1 lurker ratio (almost certainly much higher, given member!=active member, and not all members visit every day)
For example: In any given month I visit HN from a laptop, desktop, and work computer using both Chrome and Firefox, in and out of incognito mode, as well as from two different browsers on my iPad and often from other people's computers. Obviously I am not always logged in to all of these browsers so depending on how pg counted uniques (my understanding is that cookies are the common practice though I could be totally off base on this) I could be counted as anywhere from 3 to 15 uniques a month.
It would be really nice to have dataset available for researchers with username, their posts and their comments. As an example, we could use the "SPEAR Algorithm"[1] to find expertise in specific domains per HN user. [1] http://www.michael-noll.com/projects/spear-algorithm/
If pg wouldn't mind me scraping the site, in a manner that would not impact the sites performance, I will write an app to collect this data and publish it to the public domain.
It sets a crawl-delay of 30 seconds.
Earlier quoted context omitted.
If pg wouldn't mind me scraping the site, in a manner that would not impact the sites performance, I will write an app to collect this data and publish it to the public domain.
Can't speak for pg, but you should probably follow the site's robots.txt file's rules: http://news.ycombinator.com/robots.txt It sets a crawl-delay of 30 seconds.
Earlier quoted context omitted.
Here's a graph I found explaining the "link wheel" strategy: http://lemonarian.com/images/majorwheel.jpg
The sad thing is that it's not just spammers who need to do this. Even people who are genuinely creating great content in an attempt to help others still need to create backlinks from social sites if they want to show up in Google. I've made all sorts of resources that are far better than anything else that currently exists, but without actively going out and building a few backlinks these pages would get literally z…
I tried searching my own username and it does not appear in the results.
Perhaps I just failed at using Google?
Earlier quoted context omitted.
If someone creates an account but never posts, will there ever be a public-facing link the the userid page? If not, then Googlebot could be doing its job perfectly but have a totally inaccurate number of registered users using this method. I'm curious what the actual number is now...
Yes, somehow. See this: http://www.google.com/#sclient=psy&hl=en&source=hp&#...