[layer8]

> The median tenure in tech is one to two years before moving on to another team or company [1].

Note that individuals who switch jobs often inherently will figure in a higher number of tenures that you’re taking the median of here. This means the majority of software engineers will actually have a longer tenure.

[eckza]

Jobs Georg, who has changed jobs 10,000 times in the last three years, is an outlier and should not have been counted.

[philipov]

So what we should be looking at is the median number of tenures, not the median length of tenures? How would one weight the latter by the former to get a more robust measure of tenure length?

[layer8]

Well, what do you want to ascertain? In any case you won’t capture the long tail of long tenures, because you don’t know how long current tenures will extend into the future. What you could measure is the median current seniority of tech employees (how long they are in their current job), which would be a lower bound for the median of (eventual) tenures of current employments.

[twic]

Depends how you sample. If you look at all current employees, and ask how long they were in their previous job, i don't think it does.

[layer8]

That will also skew towards shorter tenures, (a) because shorter tenures tend to occur earlier in a career rather than later (so a previous tenure is more likely to have been shorter), and (b) when you’re sampling at a random point within a career, you’re more likely to hit a long tenure within that career, which conversely means that the previous tenure (or more generally, any of the past or future tenures) will more likely be shorter than the current one in that career.