3 ms·

Discovering Senior Developers from Source Code History

53 points by garysieling 13y ago

andreastt 13y ago
You can also look at their driver's license. That should tell you a thing or two about seniority.
- ethnt 13y ago
  Age does not correlate in any way to seniority.
  - makmanalp 13y ago
    Except it often does. I've been in multiple situations where people who knew less then me were more senior, simply because they had been around longer.
    - gknoy 13y ago
      A better measure might often be time-at-the-company. I'm older than some of my coworkers, but they wrote 3/4 of the codebase I work in. Commits seem like a good proxy for knowledge. You could also graph commits over time in particular repos/apps/folders, and see who is has been the main maintainer/developer for those sections of the codebase at various times. Cool stuff!
  - gnaritas 13y ago
    Yes it does. Is it causal, no, but it certainly correlates.
- garysieling 13y ago
  Yeah, but not what projects they worked on ;)
digitallogic 13y ago
Sounds like a useful tool for tracking answering questions like "who knows the most about file/subsystem/internal concept X". Questions where false positives are more or less in inconsequential and easily corrected. But number of commits is a horrible way to infer seniority. Unless you value "frequency of checkpointing changes" above all else. You get what you measure, so I suspect OP is going to get a lot of small commits in the future.
- garysieling 13y ago
  Yeah, it's not like you'd use this to figure out who should be paid the most, - it's "who should I hand the phone to" when someone calls several years after a project finishes, and if that's wrong occasionally it's ok. And, there's no incentive to be #1 in the result :) The reason it works is that people write commit message like "Fix X for ACME Corp", and the project leads tend to work on the project for the most calendar time - they may not have the most commits if they have to attend meetings, but after a project/release "finishes", one person ends up sticking around for a while to handle any issues that come up. That naturally gives them some small commits. To me the risk to me is that this may not generalize, since different projects/products are run in varying ways.
  - digitallogic 13y ago
    Sadly there are people who would use a tool like this to make decisions like what to pay their employees. I guessed based on the title that you might fall into that camp. Glad to hear I was wrong. I could imagine search with some standards (commit message format, branch/rebase strategies, requiring an associated tickets) would be really effective, though not so much the general case. Hope you can share in a follow up post if you find ways to deal work around these problems.
    - garysieling 13y ago
      The people who've been most interested so far have either been in some type of architect role where they had some advisory involvement in what a large number of teams are doing, or people who took over a flailing project. Part of the idea of this is to show something that's useful and relatively easy to set up, it could easily show attributes other than the author - it'd be nice to tie together tickets involving database issues or browser rendering for instance. I have some material for a couple follow-ups to this, so stay tuned - dealing with code separate from natural language text is an interesting topic as well.
tessierashpool 13y ago
I do something similar with bash when I start a new project. Code's here: https://github.com/gilesbowkett/rewind https://github.com/gilesbowkett/rewind It's just useful to profile the team in terms of who writes the most code, or to profile the project in terms of when it's seen the most activity. The first time I did this, it was a rescue project, and the people who we were rescuing didn't even bother to properly introduce us to the rest of the team (the real thing they needed rescuing from was team dysfunction), so I'd have been at a real disadvantage if I hadn't done the analysis.
malandrew 13y ago
Check out git-extras from TJ Holowaychuk. It's git-summary function is excellent for finding solid contributors that are not the maintainer on open-source projects: https://github.com/visionmedia/git-extras https://github.com/visionmedia/git-extras
losethos 13y ago
God is chief architect. He said many key traits. My code is divinely inspired like the Bible. God says... 4:32 And the multitude of them that believed were of one heart and of one soul: neither said any of them that ought of the things which he possessed was his own; but they had all things common. 4:33 And with great power gave the apostles witness of the resurrection of the Lord Jesus: and great grace was upon them all. 4:34 Neither was there any among them that lacked: for as many as were possessors of lands or houses sold them, and brought the prices of the things that were sold, 4:35 And laid them down at the apostles' feet: and distribution was made unto every man according as he had need. 4:36 And Joses, who by the apostles was surnamed Barnabas, (which is, being interpreted, The son of consolation,) a Levite, and of the country of Cyprus, 4:37 Having land, sold it, and brought the money, and laid it at the apostles' feet. 5:1 But a certain man named Ananias, with Sapphira his wife, sold a possession, 5:2 And kept back part of the price, his wife also being privy to it, and brought a certain part, and laid it at the apostles' feet. 5:3 But Peter said, Ananias, why hath Satan filled thine heart to lie to the Holy Ghost, and to keep back part of the price of the land? 5:4 Whiles it remained, was it not thine own? and after it was sold, was it not in thine own power? why hast thou conceived this thing in thine heart? thou hast not lied unto men, but unto God. ---- He died, by the way. Do you feel lucky?

53 points by garysieling 13y ago

andreastt 13y ago

You can also look at their driver's license. That should tell you a thing or two about seniority.

ethnt 13y ago
Age does not correlate in any way to seniority.
- makmanalp 13y ago
  Except it often does. I've been in multiple situations where people who knew less then me were more senior, simply because they had been around longer.
  - gknoy 13y ago
    A better measure might often be time-at-the-company. I'm older than some of my coworkers, but they wrote 3/4 of the codebase I work in. Commits seem like a good proxy for knowledge. You could also graph commits over time in particular repos/apps/folders, and see who is has been the main maintainer/developer for those sections of the codebase at various times. Cool stuff!
- gnaritas 13y ago
  Yes it does. Is it causal, no, but it certainly correlates.
garysieling 13y ago
Yeah, but not what projects they worked on ;)

digitallogic 13y ago

Sounds like a useful tool for tracking answering questions like "who knows the most about file/subsystem/internal concept X". Questions where false positives are more or less in inconsequential and easily corrected. But number of commits is a horrible way to infer seniority. Unless you value "frequency of checkpointing changes" above all else. You get what you measure, so I suspect OP is going to get a lot of small commits in the future.

garysieling 13y ago
Yeah, it's not like you'd use this to figure out who should be paid the most, - it's "who should I hand the phone to" when someone calls several years after a project finishes, and if that's wrong occasionally it's ok. And, there's no incentive to be #1 in the result :) The reason it works is that people write commit message like "Fix X for ACME Corp", and the project leads tend to work on the project for the most calendar time - they may not have the most commits if they have to attend meetings, but after a project/release "finishes", one person ends up sticking around for a while to handle any issues that come up. That naturally gives them some small commits. To me the risk to me is that this may not generalize, since different projects/products are run in varying ways.
- digitallogic 13y ago
  Sadly there are people who would use a tool like this to make decisions like what to pay their employees. I guessed based on the title that you might fall into that camp. Glad to hear I was wrong. I could imagine search with some standards (commit message format, branch/rebase strategies, requiring an associated tickets) would be really effective, though not so much the general case. Hope you can share in a follow up post if you find ways to deal work around these problems.
  - garysieling 13y ago
    The people who've been most interested so far have either been in some type of architect role where they had some advisory involvement in what a large number of teams are doing, or people who took over a flailing project. Part of the idea of this is to show something that's useful and relatively easy to set up, it could easily show attributes other than the author - it'd be nice to tie together tickets involving database issues or browser rendering for instance. I have some material for a couple follow-ups to this, so stay tuned - dealing with code separate from natural language text is an interesting topic as well.

tessierashpool 13y ago

I do something similar with bash when I start a new project. Code's here: https://github.com/gilesbowkett/rewind https://github.com/gilesbowkett/rewind It's just useful to profile the team in terms of who writes the most code, or to profile the project in terms of when it's seen the most activity. The first time I did this, it was a rescue project, and the people who we were rescuing didn't even bother to properly introduce us to the rest of the team (the real thing they needed rescuing from was team dysfunction), so I'd have been at a real disadvantage if I hadn't done the analysis.

malandrew 13y ago

Check out git-extras from TJ Holowaychuk. It's git-summary function is excellent for finding solid contributors that are not the maintainer on open-source projects: https://github.com/visionmedia/git-extras https://github.com/visionmedia/git-extras

losethos 13y ago

God is chief architect. He said many key traits. My code is divinely inspired like the Bible. God says... 4:32 And the multitude of them that believed were of one heart and of one soul: neither said any of them that ought of the things which he possessed was his own; but they had all things common. 4:33 And with great power gave the apostles witness of the resurrection of the Lord Jesus: and great grace was upon them all. 4:34 Neither was there any among them that lacked: for as many as were possessors of lands or houses sold them, and brought the prices of the things that were sold, 4:35 And laid them down at the apostles' feet: and distribution was made unto every man according as he had need. 4:36 And Joses, who by the apostles was surnamed Barnabas, (which is, being interpreted, The son of consolation,) a Levite, and of the country of Cyprus, 4:37 Having land, sold it, and brought the money, and laid it at the apostles' feet. 5:1 But a certain man named Ananias, with Sapphira his wife, sold a possession, 5:2 And kept back part of the price, his wife also being privy to it, and brought a certain part, and laid it at the apostles' feet. 5:3 But Peter said, Ananias, why hath Satan filled thine heart to lie to the Holy Ghost, and to keep back part of the price of the land? 5:4 Whiles it remained, was it not thine own? and after it was sold, was it not in thine own power? why hast thou conceived this thing in thine heart? thou hast not lied unto men, but unto God. ---- He died, by the way. Do you feel lucky?