The Importance of Time Capping

Time cap­ping can be defined as fix­ing the num­ber of hours for a cer­tain task, re­search pro­ject or de­ci­sion and keep­ing our re­search within those bounds. Most tasks can be com­pleted at differ­ent lev­els of depth and re­search it­self is never-end­ing—a sin­gle topic could of­ten be re­searched in an hour or could equally have an en­tire PhD made out of it. The same can ap­ply for web­site de­sign, out­reach, pol­ish­ing or many other tasks that an or­ga­ni­za­tion en­gages in. Given tasks that are not time capped, peo­ple will gen­er­ally spend more time on do­ing what they find fun or what they get ab­sorbed in in­stead of what is best to put hours into in the long run. By set­ting a time cap on a task we are pre-de­ter­min­ing how im­por­tant that task is rel­a­tive to other coun­ter­fac­tual tasks. This ap­proach of­ten re­sults in more get­ting done at some cost of depth, as of­ten 90% of the value of tasks is cap­tured by the first 10% of the effort .

There are a few spe­cific kinds of tasks that benefit es­pe­cially from time cap­ping. Re­search and de­ci­sion mak­ing are tasks that fre­quently re­sult in end­lessly flex­ible dead­lines with­out a clear ‘good enough’ point.


Re­search is one of the fore­most ar­eas that can benefit from time cap­ping. At Char­ity En­trepreneur­ship we al­lo­cate a spe­cific time frame for ev­ery re­search pro­ject. This has al­lowed our team to cover a lot of ground in a pre­dictable way, and to cover the ground we need for mak­ing a spe­cific de­ci­sion within a clear time pe­riod (e.g. con­duct­ing re­search for a year and then recom­mend­ing a list of top char­i­ties that could be started).

Many re­search teams could think of situ­a­tions where twice as much re­search, but with half the time put into each piece, would likely re­sult in more good for the world. For a more con­crete ex­am­ple: we had a very spe­cific time bud­get for our an­i­mal re­ports. For each of the re­ports we could have gone much deeper (we ended up spend­ing 1-5 hours per re­port and pub­lished a 1-page sum­mary for 15 an­i­mals). But it would have been at the cost of breadth: for ex­am­ple, we could have spent 2-10 hours per re­port, but only cov­ered 7 an­i­mals. Or, in the­ory, cov­ered 30 an­i­mals at half the depth. The key ques­tions to con­sider were what the pur­pose of these re­ports was and how big a role did they play in our endline goal (start­ing effec­tive char­i­ties). What level of depth would give us enough in­for­ma­tion so that we could start to com­pare and con­sider which an­i­mals would be a pri­or­ity? We de­cided that 1-5 hours would give us enough in­for­ma­tion for the soft pri­ori­ti­za­tion of which an­i­mals to fo­cus on.

I think highly in­tel­lec­tual cul­tures (such as the EA move­ment) tend to un­der­value this sort of time-capped re­search ap­proach, and I of­ten see com­ments on re­search, even very deep re­search, that more or less trans­late to “put more time into this re­search”. Of course, some­times it is true that more time should be put into a spe­cific branch of re­search, but I very rarely see it the other way around—com­ments that roughly trans­late to “you should have put less time into this re­search”. It’s be­come a cliche in aca­demic re­search to say “more re­search is needed,” but in some ar­eas this is re­ally not the case, and par­tic­u­larly so once coun­ter­fac­tu­als are taken into ac­count. Time cap­ping, I think, is the one step that can be used to im­prove the situ­a­tion. If some­one wants to make the case that we should have cov­ered 7 an­i­mals, but at dou­ble the depth, I am very in­ter­ested to hear the con­sid­er­a­tions in favour of mak­ing this trade­off, but if they say that more hours put into the re­search would make it bet­ter with­out a thought to­wards coun­ter­fac­tu­als then it’s harder to en­gage with.

De­ci­sion making

De­ci­sions are in­finitely com­plex and, much like re­search, it would be easy for a per­son to spend any­where be­tween 1 minute and mul­ti­ple years to con­sider com­plex de­ci­sions. Some de­ci­sions are suffi­ciently com­plex that the perfect an­swer can never be found; just bet­ter and worse guesses. Given the un­cer­tainty and com­plex­ity of the world, but also the im­por­tance of mak­ing many de­ci­sions (of­ten hun­dreds a day), time cap­ping the more lengthy de­ci­sions seems like an op­ti­mal solu­tion. This stops anal­y­sis paral­y­sis and con­stant in­de­ci­sive­ness. It also gives im­por­tant de­ci­sions a fixed dead­line and timeline for when they need to be made.

Us­ing an ex­am­ple within CE: at the end of the year, we ul­ti­mately have to choose what are the best char­i­ties for us to recom­mend. This de­ci­sion will al­most by defi­ni­tion always be in­com­plete and sub­ject to re­vi­sion. But like with the re­search above, the im­por­tant ques­tion is what are the coun­ter­fac­tu­als? If CE, for ex­am­ple, did 5 years of re­search and de­ci­sion mak­ing in the an­i­mal space, would it re­sult in higher im­pact char­i­ties recom­mended at the end? Al­most definitely. But we also have to con­sider the im­pact of a strong char­ity, or two, that could have been started 4 years ear­lier, as well as the cost of re­search. CE could cover and recom­mend char­i­ties of men­tal health, poverty, far fu­ture, and meta sci­ence. If 2 char­i­ties came from each of those ar­eas dur­ing those 4 years that would be 8 char­i­ties not started for the end benefit of a bet­ter an­i­mal recom­men­da­tion.

Some de­ci­sions are so com­plex that it’s eas­ier to have im­pact in mul­ti­ple fields (or dou­ble the im­pact in a sin­gle ‘best guess’ field). Our team ended up think­ing that it would be quicker and higher im­pact to run a year of CE on both an­i­mals and poverty is­sues than it would be to re­search and de­cide which one is of higher im­pact.

This ar­ti­cle has also been pub­lished at Char­i­tyEn­trepreneur­

If you want to be the first to know when we will start ac­cept­ing ap­pli­ca­tions to our in­cu­ba­tion pro­gram, sub­scribe to Char­ity En­trepreneur­ship’s mailing list.