5 ms·

I don't see why everything has to be bucketed into 'good' vs. 'bad'. A well-populated GitHub profile can be a positive signal to me. It makes it easier for me

I don't see why everything has to be bucketed into 'good' vs. 'bad'. A well-populated GitHub profile can be a positive signal to me. It makes it easier for me to see how a prospective candidate works and thinks, and provides me with a more useful and instructive set of questions to ask during an interview. It is also possible that I will discover that the candidate is not sufficiently experienced to meet the requirements of the position that they are applying for, but at least this way I don't have to try to ferret out these answers or waste their time. An unpopulated GitHub profile is a neutral signal for me, and means that I need to spend more time trying to tease out the answers that are simply given to me with a well-populated profile.

soneca 6y ago
In a competitive scenario, like multiple people applying to only one open position, a “neutral” signal versus a positive signal is effectively a negative signal.
- aaronbrethorst 6y ago
  When it comes to hiring highly qualified engineers, I would love to have the problem you are describing.
- bena 6y ago
  No. It's not. It's a neutral signal. It can't tell you anything either way. Even the github profile isn't a strong positive signal, it's a weak signal. But that's ok. That's really what we're looking for: a lot of weak positive signals. So that person had a github profile. Everything else could either be a neutral or negative signal. Compare that with someone without a github, but has every other positive signal you could ask for. I'd take non-github person over github person every single time.
tdeck 6y ago
> It makes it easier for me to see how a prospective candidate works and thinks Be careful making even these assumptions. My GitHub is a subset of my side projects combined with some old work stuff that was reprioritized. Many of my repos lack tests because they're for such a small number of users it's more efficient just to test manually and handle issues personally. That doesn't mean I don't write tests in my professional life. Similarly, many projects may be older and my habits have evolved since then.
- alexpetralia 6y ago
  I don't think he's making assumptions. Is it unreasonable to say "more data allows me to make a better decision?" It could be completely noise! But it is still data. I think in fact you are making assumptions based simply on the fact that he is observing data.
  - wodenokoto 6y ago
    > Is it unreasonable to say "more data allows me to make a better decision?" Yes. If you read data about A as data about B, more data will give you a worse understanding. > It could be completely noise! But it is still data. You are not technically wrong in saying that noise is data, but your implication that it is somehow better than no noise is wrong.
    - alexpetralia 6y ago
      Ah ok that make sense. As a more philosophical question, how do you know a priori if it is noise or signal? Wouldn't you have to look at it?
      - wodenokoto 6y ago
        The discussion in this thread is basically this: A says: “I like it when a candidate has a github, because it gives me and idea about how they code” B replies: “please don’t. My github code, like an unknown proportion of all github profiles, is hobby code and doesn’t follow any of the standards my professional code follows.” In this case, you need a prior that says “even if it looks legitimate, the data might be completely invalid and we cannot tell from the data”
- aaronbrethorst 6y ago
  Good thing I can flexible and ask you questions about your approach to your professional work versus your side projects.

soneca 6y ago

In a competitive scenario, like multiple people applying to only one open position, a “neutral” signal versus a positive signal is effectively a negative signal.

aaronbrethorst 6y ago
When it comes to hiring highly qualified engineers, I would love to have the problem you are describing.
bena 6y ago
No. It's not. It's a neutral signal. It can't tell you anything either way. Even the github profile isn't a strong positive signal, it's a weak signal. But that's ok. That's really what we're looking for: a lot of weak positive signals. So that person had a github profile. Everything else could either be a neutral or negative signal. Compare that with someone without a github, but has every other positive signal you could ask for. I'd take non-github person over github person every single time.

tdeck 6y ago

> It makes it easier for me to see how a prospective candidate works and thinks Be careful making even these assumptions. My GitHub is a subset of my side projects combined with some old work stuff that was reprioritized. Many of my repos lack tests because they're for such a small number of users it's more efficient just to test manually and handle issues personally. That doesn't mean I don't write tests in my professional life. Similarly, many projects may be older and my habits have evolved since then.

alexpetralia 6y ago
I don't think he's making assumptions. Is it unreasonable to say "more data allows me to make a better decision?" It could be completely noise! But it is still data. I think in fact you are making assumptions based simply on the fact that he is observing data.
- wodenokoto 6y ago
  > Is it unreasonable to say "more data allows me to make a better decision?" Yes. If you read data about A as data about B, more data will give you a worse understanding. > It could be completely noise! But it is still data. You are not technically wrong in saying that noise is data, but your implication that it is somehow better than no noise is wrong.
  - alexpetralia 6y ago
    Ah ok that make sense. As a more philosophical question, how do you know a priori if it is noise or signal? Wouldn't you have to look at it?
    - wodenokoto 6y ago
      The discussion in this thread is basically this: A says: “I like it when a candidate has a github, because it gives me and idea about how they code” B replies: “please don’t. My github code, like an unknown proportion of all github profiles, is hobby code and doesn’t follow any of the standards my professional code follows.” In this case, you need a prior that says “even if it looks legitimate, the data might be completely invalid and we cannot tell from the data”
aaronbrethorst 6y ago
Good thing I can flexible and ask you questions about your approach to your professional work versus your side projects.