Summary
I have a sense that,
1) A meaningful number of people would be able to devote themselves more to resource-intensive EA pursuits if they had a more assured guarantee of something to fall back on if the endeavour doesn't work out
2) Finances and general long-term career progression are the most important thing put at risk by path changes, followed by social status both inside and outside of EA
3) All of these problems are tractable
My concrete prediction: providing de facto insurance of various types to people with high potential and impactful projects-to-be will increase the number of them who take up those projects and this will be especially pronounced among the low-probability high-expectation endeavours.
Epistemic status before hearing the same idea in an 80k podcast episode: This is mostly based on my personal experience, which is limited and biases me pessimistically in this context for various reasons. I expect the comments to have insight(s) which reduce my conviction in the above points, but I will describe them without caveat as to not double-count the outside view. This is also my first post for whatever that's worth epistemically.
Epistemic status afterwards: Slightly more confident now, this is a good idea!
1 Financial risks
To start with the obvious, people need to put food on the table and quitting a job is an obvious threat to most people's financial security. This consideration is probably already baked into existing grant schemes to some extent but those schemes aren't explicitly trying to solve the problem of what happens to someone after a project. They can apply for another grant but people might reasonably fear the uncertainty of relying on another grant, especially given that the failed project wouldn't then have done as much to improve their track record like a successful project would have.
2 Long-term career progression
These concerns have been mentioned publicly (see How do employers not affiliated with effective altruism regard experience at EA-affiliated organizations? ). On a personal note, it seems like a concern intuitively when I consider my own career, so that post and similar ones resonate with me. Whether or not this is a concern in your specific situation depends on many things, but it's likely that there are cases where there is a genuine trade-off for someone when they diverge from their intended path. A good solution would aim to remedy this to some extent.
Aside: I would like someone with any perspective on this to write a post of their own examining this issue at a deeper level - there is a big opportunity for addressing these fears directly and carving out some model for thinking about these compromises.
3 Social status and recognition
Social status is a significant part of people's career decisions. First, consider a given person's felt sense of social status outside of EA circles. It is generally not optimized by undertaking a project that requires a deviation from the occupation someone would have had chosen purely selfishly; the social status benefit of a career is priced into that assessment for many people. On top of that fixed loss is the potential loss that would come if the desired outcome of a project isn't reached or the person decides to quit early. This is a problem because we want people to be taking detours for good causes! Worse yet, there is probably a relative deficit of people taking up low-probability high-return (net greater-than-average expected value, see Hits-based Giving) projects and these are going to be penalized even more often.
Secondly, consider someone's social standing within EA. Luckily this is something we can do more about easily and community members are generally charitable enough to recognize a good effort even when things don't work out. That said, it is still valuable to have some mechanism that allows for a likely second chance for someone to achieve what they believe they are capable of doing. Now, people do vary in the weight they place on social status or standing but those who don't care as much for it still might appreciate recognition and respect for doing good things that are personally costly. Solutions should be aimed at addressing this as well.
4 Possible solutions
4.1 Direct help within local EA networks
Helping people get back on their feet with a job, either inside or outside of EA is a good use of connections and networks that people in EA groups might have. As much as some people don't enjoy the "networking for career advancement thing", doing so within EA circles for the purpose of helping someone have backup job options is exactly in the spirit of this idea. Going even further, assuring someone you trust that (within reason) you'll be a reference or connection in their job search after a project can go a long way. Social status needs outside of EA are best met by this if used for jobs outside of EA - imagine someone being able to say "well, people aren't sure about this global health project, but if it doesn't work out my friend will help me get back into my regular career and I can continue earning to give until something else comes along".
4.2 Transition grants
This is suggested in the podcast episode. A transition grant would be what it sounds like: a grant given to someone either at the end of a project or at a point where it would be good for them to stop it early, with the aim of leading them into another project. To best address the problems I outline above, initial project grants could come with an option for a transition grant to be exercised later if needed.
4.3 Honouring ex ante good decisions
This is also suggested in the podcast episode. This could be as simple as a section in the Wiki or a dedicated website where someone at least gets their name on the wall, in a way. This goes towards improving the recognition people receive for taking chances on things with large upsides even when they don't work out.
5 Some objections
There could be ways of people abusing a transition grant systems but I don't see it as too likely and it can probably be mitigated once we have an idea of the exact structure that could be taken on. That said, this is a crux for my support of providing guaranteed transition grants in advance and some red-team thinking on this would be productive. If guaranteed transition grants aren't workable, they can still be given ad hoc by granters (as I assume was being suggested in the 80k podcast) and serve most of the same purpose. They would still be enabling some of the benefits to the general perception of the community having your back if things don't work out, just indirectly.
A reduced pressure to succeed might have negative nth-order effects, but I don't think this is a net downside. Though some people do thrive in high-stress environments, the day-to-day pressure to make something work once the 'boats have been burned' is probably bad for a group's environment and culture. See any recent scientific lab scandal for examples of what happens when the pressure for results coming either from the lab's leadership or the general academic culture boils over. Separately, though some people do thrive under higher-stress environments, this is impairing to other people. See Yerkes-Dodson law, bidirectional effects of stress on cognitive performance by COMT genotype etc.
The main crux for everyone, of course, will be how effective it is relative to existing granting schemes, which is an empirical question that is beyond my abilities to answer currently.
6 Conclusion
Making effective work as viable and compelling as possible is essential if we hope to grow EA. Underwriting some of the risks associated with taking on a project is a good way to bolster participation in the work we want people to be doing and is something that organizations which are trying to change behaviour at scale should be considering.
I would echo that "career insurance" would be particularly valuable to certain groups. Moreover, the negative effects of its absence are not uniform -- they disproportionately affect certain types of endeavors and thus warp the types of new initiatives that get off the ground (not merely who gets them off the ground).
Absence / paucity of career insurance disproportionately affects:
(Many of these points are referenced in the original post.)
Some of those disproportionate effects apply to individuals, but I submit that there is a decent correlation between the characteristics of the potential worker and the types of work they would be most interested in / good at. Moreover, some of the effects relate directly to the nature of the work.
Great examples there, thank you for commenting!
And I agree with the other point as well- it's a one-two punch in the sense of the lack of safety net pushing away certain groups of people and thereby also biasing the type of work done away from what would be otherwise optimal.