A couple tips that seemed to help me:
I agree that option value is important, but I think there's a trap where preserving option value means never testing one path. I lean toward trying to rapidly and cheaply test multiple paths, while preserving option value.
Thanks for the comment, Meerpirat. This is the latter, but felt closely enough related to use the same terminology. I'd started writing the "Getting Excited about Efficiency" post and realized that the idea didn't resonate with some people because they didn't viscerally grok why getting more stuff done was valuable. So I wrote this post about why people should care about the ideas in Half-Assing It, or my later Noticing and Getting Excited posts.
I find it useful to stagger asking for advice, roughly from easy to hard to access. E.g. if I can casually chat with a housemate about a decision when I just need a sounding board, I'll start there. Once I have more developed ideas, I'll reach out to the harder to access people, e.g. experts on the topic or more senior people who I don’t want to bother with lots of questions.
So, that example looks like an example of time pressure, rather than just being aware of time.
My understanding is that the literature on time pressure is considerably more nuanced and interesting. At its simplest, increased pressure (e.g. tight deadlines or expectation of evaluation) seem to improve performance on tasks where it’s clear exactly what needs to be done. On tasks that require creativity or novel problem solving, pressure seems to reduce performance compared to low to moderate time pressure. E.g. Ted Talk and study. I haven’t actually looked at this since college, so I can send you the dozen or so other papers I read then if you want to look at it with fresh eyes.
From that, I would expect your concern to be accurate only some of the time, albeit for some important work.
On the other hand, I have several anecdotal data points that regular time tracking is valuable for improving prioritization, though I expect the return is more varied than for short periods of time. I expect time tracking to be extremely valuable for short time spans (about 2 weeks) as a sanity check/improving knowledge of where time is spent.
Additionally, I expect people to be pretty bad at estimating productive time without tracking their time, hence the concern that prompted my original comment. The data means less if people are highly inaccurate when estimating time.
Last year, I looked at some studies to try understanding how correlated self-reported and objective measures are. There was a wide variance, with generally low to moderate correlations. When I looked just at the couple data points that are easily and/or frequently measured, the correlation was much higher, above r=0.7. Things that aren’t frequently measured have average correlations closer to r=0.3. Here’s that data if you want to reexamine it:
For numbers that were not frequently measured, the correlation between self-reported and directly measured was moderate: for one meta-analysis on physical activity, the mean r coefficient = 0.37 (range -0.71 to 0.96); for various measures of ability, mean r = 0.29 (range -0.6 to 0.80); for sedentary time, r<0.31; for physical activity, r=0.11.
A few more studies reported r coefficient ranges, but not mean r: for another measure of sedentary time, the coefficients ranged from 0.02 to 0.36; for another study on physical activity, the coefficients ranged from 0.46 to 0.53 (p value did not meet .05 threshold); for various other measures of sedentary time, the coefficients ranged from 0.50 to 0.65. If these are included in the above graph, the mean R goes up closer to .33.
For numbers that are frequently measured, the correlation between self-reported and directly measured was noticeably higher: for course grades, median r = .76 (range.70 to .84); for height and weight, median r = .94 (range .90 and above). This mildly sketchy unpublished review of hundreds of comparisons found an average of 85% perfect match between self-reports and objective records. The examples they give (e.g. self-report of hospitalizations or how many ambulatory physician visits compared with medical records) range from 89% to 100% exact match, and are mostly more frequently/easily measured.
Do you know what the landscape is of people working on this now, and whether any of them are doing it in an EA-ish way?
The biggest expenses are costs typically paid by the employer separately from salary (e.g. self-employment taxes and health insurance together are about $16,000). The next largest is outsourcing some work to help me scale coaching.
This question is too broad for me to fully answer, but checking out the productivity tips on my fb page and reading Deep Work are probably decent places to start.
I wrote up some advice for people interested in becoming coaches a while ago, you can check it out here.