I, like most humans, am bad at understanding randomness and good at spotting patterns that don’t necessarily exist. I also frequently have thoughts like: “That’s the third time this paper has been rejected. It must be bad.” These things are related.

When I submit my work, all of the variables at play, including the quality of the thing being judged, combine to give me a probability that a positive outcome will occur e.g. 0.4 – 2 out of 5 times, a good thing will happen. BUT, probabilities produce lumpy strings of outcomes. That is, good and bad outcomes will appear to us pattern-spotting humans to be clustered, rather than what we would describe as “random”, which we tend to think of as evenly spaced (see the first link above).

To illustrate, I did something very straightforward in Excel to very crudely simulate trying to publish 8 papers.

Column A: =RAND() << (pseudo)randomly assign a number between 0 and 1; in the next

Column B: =IF(Ax>0.4, 0,1) << if the number column A (row x) exceeds .4, this cell will equal 0, otherwise it will equal 1.

Thus, column B will give me a list of successes (1s) and failures (Os) with an overall success rate of ~.4. It took me four refreshes before I got the following:

Although the rejections look clustered, they are all independently determined. I have almost certainly had strings of rejections like those shown above. The only thing that has made them bearable is that I have switched papers, moving on to a new project after ~3 rejections, at the same time giving up on the thrice-rejected paper I assume to be a total failure. As a result, I am almost certainly sitting on good data that has been tainted by bad luck.

Stick with it. It evens out in the end.