3 ms·

That's not really correct, if they'd lost 100% productivity instead then each person wouldn't have been (75 - 0) / 3 = 25% less productive. Dividing like that's

• fhars 18y ago
  Indeed. They should have produced e.g. 400 productivity units (lines of code <gd&r>?), but lost 30%, i.e. 120 units. 100 are due to the bad apple, the remainig due to lost productivity of the other three. So each if them only produced 93.333 units, for a loss of 6.667% of individual productivity. Which is 233% more than 2%.
  • ajb 18y ago
    Yes. But isn't this just a corollary of a (small) group being more than the sum of its parts? IE, A group of 4 should be more than 4/3 productive than a group of 3. The right comparison is to compare a group of 4 with one bad apple with the productivity of a group of 3.
    • fhars 18y ago
      Yeah, the truth is somewhere between that and the other theory that the overhead for internal communication goes at least quadratic with the number of group members, so a group of four looses 16/9 more of their productivity to internal friction than a group of three. And all this is assuming you actually have a quantitative measure of programmer productivity that is accurate enough to measure a 30% change. But let's no go down that drain again.
    • anamax 18y ago
      > A group of 4 should be more than 4/3 productive than a group of 3. Why? Amdahl's law suggests otherwise, as does any non-zero cost of communications.
      • ajb 18y ago
        Because the knowledge and skill set of the group is larger. However, yes, at some point this is outweighed by the costs.