3 ms·

Lets say your boss gives you a list of ten name/email address pairs every morning at 10:00, and you have to give him a list of the three most professional candi

• angusgr 16y ago
  I'd say that this is still meaningless without context. Who is your boss? Who are you? Who are the potential hires for? Where did the names come from in the first place? If I'm reading your reply correctly, you're actually looking to write an algorithm for an automated "find-a-hacker" engine? In which case, I'd say the first bar to surpass is from another comment here - the guy who said his website is pretty crummy but it comes up first when you google "PHP programmer edinburgh"
  • techbio 16y ago
    Meaningless is a little strong. I'm asking generally. And clearly Google is not tuned to answer this.
  • techbio 16y ago
    I do not know your name, but for example, projectgus gives me 3400 Google results: blog, twitter, github, and then the rest of the top ten are "something something Project - Gus Winters, something, ..." So 3400 is a poor rating metric. But what metrics from the list of HN, StackOverflow, /., reddit/prog, yahoo answers, MIT Media Lab, web site PR, Citseer mentions, etc. could constitute a good ranking? Sure, a Nobel Prize is a shortcut. I am talking about a proxy or data model for finding and sorting. If it can be automated, it will be.
    • angusgr 16y ago
      It would still seem to me like it would be context sensitive, in the sense that you want different information to hire different types of professionals and also that people put different aspects of their life online. To follow on from your example of me - googling projectgus gives you my both my personal blog and my twitter account as the first two results. My twitter account title gives you my real name, and if you google that then you immediately find my current employer, and some details from an old employer, and then a long tail of quasi-related noise. Whether or not you'll find anything relevant to what you want depends on what you're looking for. I don't have a resume posted online, which means that although I'm a professional programmer most of my programming experience doesn't immediately show up anywhere on the internet. However, it's possible you could infer a lot of it from my posts on various little mailing lists and forums (for example, a handful of posts on Erlang-Questions because I spent about a year writing Erlang and a few posts on the Hibernate forums because I've done a little bit of Java middleware development.) Maybe. Of course, nobody just says "let's see if Angus is a hacker we can hire". Someone might say "oh, you need more Erlang programmers? I know a local programmer called Angus who has programmed in Erlang" (for contrived example). In which case, you google my name + Erlang and you can see that I've programmed in Erlang. Although my online Erlang presence is, in my opinion, underwhelming and you would not decide to approach me on that basis. Which, in my opinion, would be a mistake. ;) However, even though you wouldn't get specific information about me, you would be able to read my blog and find out I like Linux & open source, I own an Arduino, I like taking apart cheap chinese gadgets, and I like writing. Which is possibly all useful information, again depending on exactly what you were after me for in the first place. Although I didn't put that information online to get jobs (if I did want to do that, I'd have a resume section online or a LinkedIn profile or something, and maybe the above paragraphs of rant would be redundant.) I hope this discussion helps you in whatever process you decide to automate. I understand that you're talking about something a bit different, but at the same time I think it would be hard to come up with a one-size-fits-all approach that was entirely based on scraping other data. Finding something more specific, in the same vein as Haskellers.com except without the user-submitted data, might be plausible. Stack Overflow Careers, for instance, uses Stack Overflow as its data source but OTOH it still relies on the user to submit other details, in order to make the information relevant to employers.
      • angusgr 16y ago
        Also worth noting that this process of "Google stalking" me only works because I have a relatively unique name. Doing it by email address might work, although most people make an effort to keep their email addresses away from automated harvesting. :).
        • techbio 16y ago
          I just googled the name of your website. I hope you don't actually think that is stalking.
          • angusgr 16y ago
            Of course not. I put it in air quotes because I couldn't think of a better word for "using google repeatedly to find out everything you can about someone" (referring to the process that I described, ie more involved than simply googling me.)
        • techbio 16y ago
          With Erlang, I think you could be identified easily. At some level of detail everyone is, and if I needed an experienced Erlang programmer, I would probably know where to look. But how about with PHP instead?
        • techbio 16y ago
          With Erlang, I think you could be identified easily. At some level of detail everyone is, and if I needed an experienced Erlang programmer, I would probably know where to look. But how about with PHP or C instead?
      • techbio 16y ago
        One heuristic is to connect a person's name to other identifiers, such as email addresses, phone number, web domain, whois, unique user name at any popular site, university affiliation, etc. Then search for each one, and perform some map/reduce on the results for each search. I will take a look at Stack Overflow Careers, btw.