1. 9
  1. 18

    The headline encompasses too much to be supported by the study on one specific task. It should be more like “Do years of experience make you review code faster in conditions that make you suspect it’s a trick question?” and the answer is “A bit, but not as much as specializing in code reviews”.

    1. 10

      Maybe Leetcode-style problems don’t correlate with actual job performance.

      Yeah, this little study of 37 programmers has little to no relation to what software engineering is.

      I would expect a senior engineer to be slower at reading and writing code than a junior engineer. A senior developer should immediately see first order effects and analyze second or third order effects. Juniors analyze first order effects.

      1. 5

        The post assumes all there is to programming is writing algorithms. As you grow in your career, you may be less used to writing specific algorithms to solve specific problems. But you will get experience in building large and complex systems, which IMHO, is more important than writing 10 lines of code quickly.

        1. 1

          The study is about comprehending code, and I would expect senior people to still be doing code review for more junior ones because that’s one of the best ways of transferring knowledge. That said, if I had to do code review of the snippet from the article, I’d likely bounce it back asking for some comments and then do code review once I could compare what the code was doing with what the author thought the code was doing.

        2. 4

          It looks like they had 15+ measurements but didn’t do any multiple hypothesis corrections on RQ3. If you do a quick Bonferroni¹ then none of their measures pass the significance threshold, which suggests the experiment wasn’t well set up to measure RQ3 at all.

          ¹ Bonferroni is for p-values, I don’t know if it can be applied to Spearman too? Still.

          1. 3

            I was going to reply with a similar response to @dmathieu but I realised, it’s actually true. The crux is that what we do as a job (roughly) isn’t programming. It’s communicating, collaborating, architecting, thinking ahead, etc.

            Experience may not make you a better programmer but it’ll definitely make you a better teammate.

            1. 0

              Very well put! Couldn’t agree more with this.

            2. 2

              The test seems oversimplified to the point of meaninglessness to me. Of course if you test on a task that a junior programmer can do fine, there will be no improvement with experience. In reality, the correct “senior” code review of the code in the article takes five seconds: “delete this — why on earth did you write your own substring search function?” If you tested on, say, a pull request for PostgreSQL, I bet you’d see a much stronger effect of experience. But that would be a lot harder to test.

              1. 1

                Does experience make anyone a better anything?

                1. 2

                  Is this really a question? I am better at front end development than any of the junior developers that have come into the company. Whether or not I will be better than them after all of us have five more years of experience is another question. But having some experience rather than none, the answer is certainly yes.

                  1. 1

                    It’s an important question in general because experience provides opportunities for learning, it doesn’t guarantee learning. Some people gain experience by making the same mistakes repeatedly and being surprised each time, others learn from the first mistake and don’t make it again. Most people do both in different contexts.