4 ms·

Not everybody is underpaid, but among the underpaid, they found men as well as women. As random illustration, they might pay 200k+-5% to 80% of all level X deve

by wsy 8y ago

Not everybody is underpaid, but among the underpaid, they found men as well as women. As random illustration, they might pay 200k+-5% to 80% of all level X developers, and the other 20% might only earn 150k+-5%. I still agree that 'underpaid' is a difficult term here. Maybe the 80% are overpaid? The only hard fact in this example would be that the salary is more skewed than it should be for a 'fair' compensation.

CydeWeys 8y ago
What does the ideal salary distribution curve look like, anyway? Should it be a bell curve? A long-tail distribution? Should it have distinct peaks, i.e. everyone with the same YOE is paid exactly the same? Should the salary distribution mirror the talent distribution, or no?
- wsy 8y ago
  Good questions. I think that compensation should mirror the value an employee provides to the company. However, in many cases, that is very hard or even impossible to measure. In large companies with many career levels, the role level indicates the perceived value for the company. Then, the distribution within the level doesn't matter that much, as long as the difference between minimum and maximum pay is relatively small, and there is not much overlap to the neighborhood levels. Of course, there is no objective criterion what "relatively small" means in this context. And, as the article already points out, this only covers fairness within the same level. Maybe even more important is if everybody is evaluated in the same way when it comes to career progress.
sokoloff 8y ago
If you believe that developers are within a factor of 33% of productivity (defined notionally as “value created per week”), then there is excessive skew in the numbers you propose. I believe the productivity figures are at least a factor of 2, and probably a tail out to a factor of 10 or more versus the median developer (and then a left-going distribution under the median where a substaintial number of developers do not create any value at all). Under that set of beliefs, 20% making ~$150K and 80% making ~$200K is actually insufficient skew vs a “fair” system.
- wsy 8y ago
  I wrote this in the context of the article, where a Google developer with 33% more productivity would already have progressed to the next role level. At least, that is what I would assume for large companies.
  - sokoloff 8y ago
    Promotion is not centered primarily around productivity, at least not beyond SWE2 or whatever the second engineering level is in a given company. Ability to lead/influence at a scope larger than self is largely disjoint from individual productivity in coding.

by wsy 8y ago

CydeWeys 8y ago

What does the ideal salary distribution curve look like, anyway? Should it be a bell curve? A long-tail distribution? Should it have distinct peaks, i.e. everyone with the same YOE is paid exactly the same? Should the salary distribution mirror the talent distribution, or no?

wsy 8y ago
Good questions. I think that compensation should mirror the value an employee provides to the company. However, in many cases, that is very hard or even impossible to measure. In large companies with many career levels, the role level indicates the perceived value for the company. Then, the distribution within the level doesn't matter that much, as long as the difference between minimum and maximum pay is relatively small, and there is not much overlap to the neighborhood levels. Of course, there is no objective criterion what "relatively small" means in this context. And, as the article already points out, this only covers fairness within the same level. Maybe even more important is if everybody is evaluated in the same way when it comes to career progress.

sokoloff 8y ago

If you believe that developers are within a factor of 33% of productivity (defined notionally as “value created per week”), then there is excessive skew in the numbers you propose. I believe the productivity figures are at least a factor of 2, and probably a tail out to a factor of 10 or more versus the median developer (and then a left-going distribution under the median where a substaintial number of developers do not create any value at all). Under that set of beliefs, 20% making ~$150K and 80% making ~$200K is actually insufficient skew vs a “fair” system.

wsy 8y ago
I wrote this in the context of the article, where a Google developer with 33% more productivity would already have progressed to the next role level. At least, that is what I would assume for large companies.
- sokoloff 8y ago
  Promotion is not centered primarily around productivity, at least not beyond SWE2 or whatever the second engineering level is in a given company. Ability to lead/influence at a scope larger than self is largely disjoint from individual productivity in coding.