where most of the time invested in doing something, looks like all the time you invested, will end up having a poor ROI
pretty much every point there except the end looks like a case study for calling it a sunk cost and quitting to do something else.
Its a hard thing to know when your sample size of data is big enough to make conclusions that have consequences on
(I guess all in all that's the point, you need to use some heuristic, and the link is arguing for a heuristic of 50 pgs when reading a book, which seems like a rule of thumb that I have no particular objection to)
For me it would've been 1 year into my (ultimately failed) Ph. D project. I would be a lot wealthier now if I'd launched into the tech world then and been working all that time.
There's an algorithm for this: if the risk-adjusted net present value (whole utility curve, not money) is lower than something else you could be doing, then do that instead. Obviously, humans like variety so one needs to consider goods and experiences where the commitment is in multiple units and baskets.
To bolster confidence in the decision to drop a project, it's also important to "complete" (feel like you've completed...) a certain number of projects. Wellness/self-esteem are also improved with project completion.
What's actually hard is calculating those values with any degree of confidence
If there's going to be a hockey stick growth at the end, the net present value looks a lot different, but what's my basis for knowing that it will be there?
When to quit... great question! Seth Godin wrote a wonderful little book on this called 'The Dip: A Little Book That Teaches You When to Quit (and when to stick)'
when to quit, is a really hard problem
I'm reminded of the Hacker New link from a month or so ago about 'The emotional journey of creating anything great' https://news.ycombinator.com/item?id=17164822
where most of the time invested in doing something, looks like all the time you invested, will end up having a poor ROI
pretty much every point there except the end looks like a case study for calling it a sunk cost and quitting to do something else.
Its a hard thing to know when your sample size of data is big enough to make conclusions that have consequences on
(I guess all in all that's the point, you need to use some heuristic, and the link is arguing for a heuristic of 50 pgs when reading a book, which seems like a rule of thumb that I have no particular objection to)