Five Ways To Prioritize Better

This piece is cross-posted on my blog here.

I’m go­ing to let you in on a se­cret of pro­duc­tivity.

Those peo­ple you ad­mire, the ones who make you won­der how on earth they ac­com­plish so much? Those peo­ple might work more hours than you or be more tal­ented or more pas­sion­ate. Or they might not.

But they prob­a­bly work on bet­ter things, in bet­ter ways.

Now, be­fore you protest that that’s the same thing as be­ing tal­ented or smart or hard­work­ing, let me un­pack that claim. Work­ing on bet­ter things means they care­fully choose what’s worth car­ing about, and what they won’t give a fuck about. Work­ing in bet­ter ways means they care­fully choose to do the most im­por­tant ac­tions to ac­com­plish those goals.

In short, work­ing on bet­ter things, in bet­ter ways is … pri­ori­ti­za­tion. Pri­ori­tiz­ing well is the com­mon thread be­hind suc­cess­ful peo­ple.

Be­cause pri­ori­tiz­ing well is so freak­ing im­por­tant. Cu­mu­la­tive good choices can mul­ti­ply your im­pact tens or hun­dreds of times. You can’t work tens or hun­dreds as many hours in a day. You can choose what to do and how to do it most effec­tively.

But pri­ori­ti­za­tion isn’t just fol­low­ing through on the ac­tions you already know you should do—though that’s im­por­tant too. Know­ing what you should do is ac­tu­ally hard. There’s a long gap be­tween want­ing to make the world brighter and know­ing how to do it. I want to ac­knowl­edge that.

So, bear with me a bit. This post is longer than usual be­cause I’m try­ing to give you a bunch of ex­am­ples to re­ally get a taste of what pri­ori­tiz­ing feels like. I think ev­ery one of the ex­am­ples will help you un­der­stand bet­ter, but, if you have a hard time fo­cus­ing for 14 pages, maybe read one tool now and come back an­other time for oth­ers?

I’ve or­ga­nized the ex­am­ples around five spe­cific con­cepts that you can use to help ap­ply the mind­set more effec­tively: The­ory of Change, Lean Tests, Bot­tle­necks, Bal­l­park Es­ti­mates, and the 8020 Rule. I’ll start each sec­tion with some con­crete sto­ries be­fore sum­ma­riz­ing the con­cept and how it helps you pri­ori­tize.

Note: The fol­low­ing vi­gnettes, apart from the ones about me and the in­ter­view with Tara, are fic­tional. How­ever, they are in­spired by mul­ti­ple real con­ver­sa­tions I’ve had. If you find your­self think­ing “this could never hap­pen”, rest as­sured that, though these ex­am­ples are fic­tional, they are based on real con­ver­sa­tions.

I. The­ory of Change

1. Ca­reer planning

Phil is tel­ling his friend Erica about his plans to switch into AI safety.

Phil: “So, I was think­ing I’d stay at my job for an­other six months. I have a bonus com­ing then, so it’s a bet­ter time to make a ca­reer switch. ”

Erica: “Huh, I’m sur­prised that your cur­rent job is the best plan for your goals. I thought you said the most im­por­tant goal was build­ing your ma­chine learn­ing skills?”

Phil: “That’s right, but I’m pick­ing up some ma­chine learn­ing on the job, so it’s also skill build­ing!”

Erica: “But you’re only spend­ing a bit of your time on ML, and it’s not like all of that learn­ing will trans­fer to the jobs you’re ap­ply­ing for any­way. Wouldn’t you learn a lot more by in­de­pen­dently study­ing some di­rectly rele­vant ML?”

Phil: “’ve got a point. I could learn a lot more if I took a month off to just study. But that feels riskier, and I might have a hard time stay­ing mo­ti­vated.”

Erica: “But also con­sider, you told me that you thought that work­ing on AI safety was sev­eral times as im­pact­ful as your cur­rent role, at least in ex­pec­ta­tion. That means that spend­ing six more months be­fore switch­ing is like wast­ing a whole year or more of im­pact.”

Phil: “Man, you’re right. I could start ap­ply­ing now so I’m ready to switch right af­ter I get my bonus. But, I don’t ac­tu­ally know if I’m ready! I don’t know if I have the skills I would need to get hired.”

Erica: “Then it sounds like you need to find out.”

Phil: “I know a cou­ple of peo­ple I could ask, and I can look at the job posts to see what skills they’re look­ing for. That should help. And if I’m not ready, I’ll prob­a­bly get a bet­ter idea ex­actly what I should study. Then I can make a timeline so I’m ready to switch as soon as I can.”

Erica: “How do you feel about that plan?”

Phil: “Pretty good. I think leav­ing my job feels re­ally risky given how un­cer­tain I am, and it was mak­ing me avoid think­ing about switch­ing. In­ves­ti­gat­ing more with­out feel­ing like I need to com­mit to leav­ing helps.”

When Phil worked back­wards from his end goal, he re­al­ized his origi­nal plan was ac­tu­ally a pretty bad plan (at least for him), be­cause it wasn’t go­ing di­rectly for the goal. De­spite a lot of un­cer­tainty, Phil’s best guess is that ap­ply­ing sooner is much bet­ter. Wait­ing would prob­a­bly mean he’d lose a few months that could have been spent on much more valuable work. But at worst, he might never make a plan that will ac­tu­ally have an im­pact un­less he stops and thinks harder—it’s re­ally easy to de­fault to the sta­tus quo, even when that’s a bad de­ci­sion.

2. Pub­lish­ing an op-ed

Elle wants to pub­lish an op-ed. But she doesn’t re­ally know how op-eds get pub­lished. So, she asks her friend Peter—who has pub­lished op-eds be­fore—about the pro­cess.

Elle: “So, I was think­ing I’d write an ar­ti­cle, send it to a bunch of news­pa­pers, and cross my fingers. What do you think?”

Peter: “Well, that’s prob­a­bly not go­ing to work out. You don’t just write an op-ed and send it to places; you need to re­ally tai­lor the piece to the mag­a­z­ine you want to pub­lish it in.”

Elle: “Huh, how would I do that?”

Peter: “First, you should check out other pieces on the plat­form you’re in­ter­ested in. If a venue has already pub­lished sev­eral pieces on similar top­ics, then it’s likely that an ed­i­tor there likes that topic and is more likely to ac­cept a new per­spec­tive on the is­sue. Se­cond…”

Elle lacked an ac­cu­rate model of how her ac­tions would lead to her goal be­cause she didn’t un­der­stand the world well enough. Creat­ing an ac­cu­rate the­ory of change re­quired learn­ing what ac­tions would gen­er­ally suffice to ac­com­plish her goal. If she hadn’t learned how the pro­cess ac­tu­ally worked, she might have wasted a lot of effort with­out ever get­ting the piece pub­lished. But she had no way of know­ing that un­til she asked.

3. Study­ing for the GRE

Like many stu­dents be­fore me, I didn’t re­ally care if I re­mem­bered the con­tent af­ter the test. I just wanted to spend as lit­tle time as pos­si­ble to get a GRE score I was happy with. But how to do that?

Ac­cord­ing to test prep web­sites, I should have worked through a GRE prep book (prefer­ably theirs). After all, those are ex­plic­itly de­signed to pre­pare you for the GRE. They in­clude ev­ery­thing I needed to know; con­cept re­view, vo­cab­u­lary quizzes, and prac­tice tests.

But I knew they would ac­tu­ally waste time.

See, I knew that spread­ing my time evenly over the ma­te­rial was an in­effi­cient way to learn. I’ll learn a lot more study­ing stuff I don’t know yet, rather than re­view­ing stuff I already knew. Nei­ther the study guide nor the vo­cab would have fo­cused me on just my weak­est ar­eas, so I would have spent hours reread­ing fa­mil­iar ma­te­rial.

In­stead, I op­ti­mized my study for rapidly im­prov­ing my weak­nesses: take prac­tice quant sec­tion tests (my weak­est sec­tion), study the ques­tions I missed, sum­ma­rize the con­cepts be­hind the ques­tions, re­peat. Those three steps saved me dozens of hours.

The­ory of Change

In each of these ex­am­ples, the per­son wanted to ac­com­plish a par­tic­u­lar goal. They worked back­wards from that goal to find the steps that would make them likely to suc­ceed at the goal.

Elle figured out how the pub­lish­ing world worked so that she knew what ac­tions were likely to suc­ceed. Phil found that a differ­ent path would al­low him to have an im­pact sooner and re­duce his chances of failing. I worked back­wards to cut out un­nec­es­sary work and save time.

Each per­son needed to know their goal, figure out what steps would re­li­ably lead to that goal, and then fo­cus only on those steps. That’s the the­ory of change—this model of the causal chain of ac­tions that lead to suc­cess­fully ac­com­plish­ing the end goal. You might need to in­ves­ti­gate how the world works or go learn some­thing if you don’t have enough in­for­ma­tion, in or­der to ac­cu­rately pick what will have an im­pact.

So, if you want to ap­ply this tool, ask your­self—what steps will ac­tu­ally make you likely to achieve your goals?

(If you want more, here’s a great post on the­ory of change.)

Work­ing out your goals and your the­ory of change is a pre­req­ui­site step to other pri­ori­ti­za­tion tech­niques. For ex­am­ple, it’s go­ing to be hard to use the next con­cept (lean tests) with­out hav­ing a least a start­ing point for what you want to op­ti­mize.

II. Lean Tests

1. Start­ing a business

The sum­mer af­ter my fresh­man year of col­lege, I was hired to start a com­pany mak­ing per­sonal bi­ogra­phies.

I im­me­di­ately started find­ing con­trac­tors to cre­ate the books, get­ting pro­to­types made, and build­ing a web­site. This seemed rea­son­able to me then – af­ter all, that was my job de­scrip­tion. And I did a good job. At the end of the sum­mer, I had a full pro­duc­tion plan ready for when the first cus­tomer pur­chased.

Ex­cept, no one ever pur­chased a sin­gle book.

If I had started by talk­ing with po­ten­tial cus­tomers, I could have known in a month that the idea was doomed from the start. The tar­get au­di­ence had no in­ter­est in the elab­o­rate $10,000 product my em­ployer en­vi­sioned. But I didn’t know that, be­cause I cre­ated a product be­fore I con­firmed peo­ple were ready to buy it. I could have saved an en­tire sum­mer of work if I had just tried talk­ing to po­ten­tial cus­tomers first.

2. Con­duct­ing research

Max is ask­ing his friend Ellen for ad­vice about his re­search pro­ject.

Max: “Aw man, I’m su­per bummed—I spent six months re­search­ing this policy area. Then when I sent my write-up draft for feed­back, some­one sent me an un­pub­lished doc­u­ment where they’d already done part of the re­search! Plus, now I need to do more re­search to an­swer their ques­tions. I feel like I wasted so much time already, and I’m not even done. What could I have done?”

Ellen: “That sucks. Hmm, there’s a post called Re­search as a Stochas­tic De­ci­sion Pro­cess that might help avoid similar situ­a­tions in the fu­ture. Want to hear about it?”

Max: “Yeah!”

Ellen: “The re­ally sim­ple ver­sion is to first do the parts of the task that are most likely to fail or change what other steps you will do, rather than do­ing the eas­iest parts first. This way, you re­duce un­cer­tainty about which tasks are nec­es­sary as quickly as pos­si­ble. For ex­am­ple, if your task has three steps and one is most likely to fail, you save time in ex­pec­ta­tion by do­ing that one first, be­cause you might not need to do the other steps at all.”

Max: “Yeah… I did the easy tasks first here. Like, do­ing all the re­search was a lot of work, but it was easy to just keep read­ing. Ask­ing for feed­back didn’t take much time, but it was hard. I got re­ally anx­ious when­ever I thought about send­ing the emails, and just kept do­ing more re­search, and then more re­search. But hav­ing that feed­back ear­lier would have to­tally changed what things I choose to re­search. It could have saved me hun­dreds of hours.”

3. Learn­ing a new subject

Alex wanted to do some in­de­pen­dent study to see if he would be a good fit for work­ing on AI safety. Based on 80,000 Hours recom­men­da­tions, he found a few promis­ing math courses that he could work through on the week­ends over a few months.

Be­fore he got started, he asked a few ac­quain­tances who worked on AI safety whether his plan made sense. They mostly thought it did, but agreed that sev­eral of the math top­ics he had planned to study weren’t im­me­di­ately valuable. He could safely skip those for now.

Spend­ing an hour send­ing those emails cut his months of study in half.

Lean Tests

In each of these ex­am­ples, the per­son wanted to effi­ciently ac­com­plish their goal. In the first two ex­am­ples, the per­son failed to test quickly, and wasted time or failed en­tirely. In the third, by quickly test­ing their idea, the per­son ex­posed flaws early so that they could rapidly cor­rect them or move on—in­creas­ing the pro­por­tion of their effort that ac­tu­ally made them suc­ceed.

I spent an en­tire sum­mer on a pro­ject that failed en­tirely be­cause I didn’t test it early enough to change it into some­thing that could suc­ceed. Max could have saved hun­dreds of hours by ask­ing for feed­back up­front. Alex did save hun­dreds of hours by check­ing his plan be­fore he im­ple­mented it.

Each per­son could have bro­ken their tasks into chunks to iter­ate on, get­ting feed­back each step of the way, rather than risk wast­ing time by in­vest­ing a lot of effort with­out feed­back.

Lean method­ol­ogy is about con­tin­u­ously do­ing small tests to check that you’re on the right track. This al­lows you to iter­a­tively mak­ing lots of cor­rec­tions that move you to­wards your goal, even when you’re not sure what is re­quired (or when you are sure but are wrong.)

By find­ing the flaws early, you can change course early, min­i­mize wasted time, and re­duce the risk of ul­ti­mate failure. Bet­ter to know a pro­ject will fail be­fore you put in months of effort that could have been spent on a pro­ject more likely to suc­ceed. You could get feed­back from more ex­pe­rienced peo­ple, your tar­get au­di­ence, or the thing it­self.

So, what is the first quick test you could cre­ate to get feed­back and iter­ate?

III. Bottlenecks

1. Anxiety

Anna is early in her jour­nal­ism ca­reer. She’s talk­ing to her part­ner Dan, a fel­low writer, about her draft of a piece on fac­tory farm­ing.

Dan: “Hey, did you pitch your idea to your su­per­vi­sor to­day like you planned?”

Anna: “No… I just got too ner­vous and couldn’t make the words come out. I think I want to keep work­ing on it be­fore I talk to her.”

Dan: “Anna, your piece is great! You’ve been work­ing on it for a year already. It’s not go­ing to get bet­ter with­out feed­back. All of your hard work won’t mat­ter un­til some­one sees it.”

Anna: “Yeah, you’re right. But I just feel so scared when I try. My chest gets tight and it feels hard to breathe.”

Dan: “Anna, maybe it’s time to con­sider talk­ing to some­one about this. What do you think?”

Anna: “I’ve ac­tu­ally been think­ing about it for a while now, and I think you’re right.”

2. Procrastination

Mary had meant to get started on her fi­nal pre­sen­ta­tion for her in­tern­ship two weeks ago. But she felt a wave of dread when­ever she started to think about it, so her thoughts slid away to some­thing less awful each time. Now the pre­sen­ta­tion is in a week, and the feel­ing has risen to panic mode. Yet she still can’t make her­self even look at her re­search.

This isn’t a new feel­ing for Mary. She’s six months over­due on a write-up from her former re­search po­si­tion. But when­ever her former ad­viser sends an email ask­ing for the re­port, a pit opens in Mary’s stom­ach.

Mary started coach­ing to tackle her per­sis­tent pro­cras­ti­na­tion. Over the next year, Mary and I worked to­gether to build up her abil­ity to make bet­ter plans, set up habits to re­duce dis­trac­tions, learn to ask for help early and of­ten, and find com­mit­ment mechanisms that work for her.

After a year of work­ing on this big bot­tle­neck, she feels con­fi­dent in do­ing her work by her dead­lines. That work was an in­vest­ment in her­self that will pay off over the rest of her ca­reer.

3. Fatigue

Some­times one is­sue will dom­i­nate ev­ery­thing else in re­gard to pro­duc­tivity, of­ten a phys­i­cal or men­tal health is­sue. In my case, it was fa­tigue. When you need to take three naps a day, you get less work done re­gard­less of what other pro­duc­tivity tools you use.

So, it was worth­while in­vest­ing a bunch of time and effort to im­prove this. I tried a pa­rade of ex­per­i­ments and tracked all the fac­tors that I thought might in­fluence my en­ergy lev­els – in­clud­ing sleep times, sleep du­ra­tion, hy­dra­tion, ex­er­cise, med­i­ca­tions, mela­tonin, do­ing a sleep study, tem­per­a­ture, naps, and nu­tri­tion. I tracked my en­ergy lev­els and a chang­ing sub­set of vari­ables for 3-6 months, then com­pared the odds ra­tios for each vari­able.

Here, a bunch of small things cu­mu­la­tively broke the bot­tle­neck, pri­mar­ily sleep du­ra­tion (which was fixed by a con­sis­tent sleep sched­ule), ex­er­cise, hy­dra­tion, and fi­nally an an­tide­pres­sant.

While I still have is­sues with fa­tigue, it’s no longer the key thing hold­ing me back.


In each of these ex­am­ples, the per­son needed to get past one bot­tle­neck that was hold­ing them back from suc­ceed­ing. Each bot­tle­neck eclipsed other tasks, even valuable tasks, un­til the prob­lem was ad­dressed.

Anna needed to work on her so­cial anx­iety be­fore her hard work would ever see the light of day. Mary found she could ac­com­plish sev­eral times as much once she got her pro­cras­ti­na­tion un­der con­trol. I had the en­ergy to start a blog af­ter I re­duced my fa­tigue.

Origi­nally, bot­tle­necks referred to the “rate-limit­ing fac­tors” that slow down the en­tire pro­duc­tion line. In our ex­am­ples, the bot­tle­necks are the tasks, be­liefs, or prob­lems that slow down or stop you from ac­com­plish­ing other tasks. Each per­son made progress by iden­ti­fy­ing what was hold­ing up other steps. Once they took care of the high-lev­er­age fac­tor, ev­ery­thing else went much faster.

So, what are the one or two things you could change about your­self or your en­vi­ron­ment to ac­com­plish twice as much?

IV. Bal­l­park Es­ti­mates

1. Choos­ing pro­jects

Will is talk­ing with his PhD su­per­vi­sor, Kate, about feel­ing over­whelmed by too many pro­jects.

Will: “I think I just need to choose one or two to fo­cus on, and put the rest on hold for now. But I want to work on all seven.”

Kate: “Well, to start with, which ones are most im­por­tant to work on?”

Will: “That’s the prob­lem; they all seem im­por­tant! Papers 1 and 2 have a good chance of be­ing pub­lished in a good jour­nal, which is im­por­tant if I want to con­tinue in academia. Paper 3 has a good col­lab­o­ra­tor, and I don’t want to let them down. And pa­pers 4 and 5 are ex­cit­ing. I think those ideas could re­ally be im­pact­ful. Arrg...I just feel over­whelmed when I think about it, like I need to do them all.”

Kate: “Okay, let’s try a thought ex­per­i­ment. How much would you pay to have each of these pro­jects mag­i­cally com­pleted? If it’s hard to think about pay­ing with your own money, how much do you think Open Phil would pay to have the pro­ject com­pleted?”

Will: “Argh, thats hard.” *15 min­utes of brain­storm­ing later* “Okay, pa­per 4 could be re­ally big if it goes well, and 2 and 3 are maybe most im­por­tant for my ca­reer. So I think I’d pay like $1000 each for pa­pers 1 and 5, $1500 for pa­pers 2 and 3, and $5000 for pa­per 4. But these are su­per crude, I’m re­ally just guess­ing here.”

Kate: “Crude num­bers are fine. You’re re­ally just try­ing to get a bet­ter sense of how you in­tu­itively value each of these. Those num­bers help clar­ify the rank­ing and rough mag­ni­tude of differ­ence be­tween the pro­jects. Sounds like pa­per 4 is the best, and then 2 and 3 are a bit bet­ter than 1 and 5, all else equal. Now, which pro­jects do you ex­pect to take the least time?”

Will: “Paper 3 for sure. My col­lab­o­ra­tor is do­ing a bunch of the work, so it’s prob­a­bly half as big as the other pa­pers. Between the oth­­ally hard to say. Um, so maybe I’d sort them 3, 1, 4, 5, 2 from least to most time, but I’m re­ally guess­ing here.”

Kate: “Based on value and time re­quired, it sounds like you should spend most of your time on pa­per 4, plus some on pa­per 3.”

Will: “But all of these are es­ti­ma­tions! I’m not con­fi­dent, and I could be re­ally wrong.”

Kate: “You’re not go­ing to be con­fi­dent. Things are un­cer­tain and will be un­cer­tain no mat­ter how long you think about it. So you have to make your best choice de­spite your un­cer­tainty. Es­ti­mates are a way to try to make that choice as well as you can. And you should ab­solutely spend more than 5 min­utes think­ing about them. So, take a few days to think about your es­ti­mates in more depth. Maybe ask a cou­ple of ad­vi­sors. But, when you’re done, go with the high­est ex­pected value and stop wor­ry­ing about it. You can change your plan if you get new in­for­ma­tion. For now, you’re do­ing the best you can, and that’s good enough.”

Will: “I...think...I can do that. Thanks, Kate.”

2. Learn­ing a new skill

Lyra is talk­ing with her coworker Mike about her plans for in­de­pen­dent study to im­prove her cod­ing skills.

Lyra: “So, I’m de­bat­ing be­tween spend­ing a big­ger chunk of time to re­ally un­der­stand com­puter ar­chi­tec­ture or do­ing sev­eral small learn­ing pro­jects around things that came up in my job. I’m hav­ing a hard time mak­ing progress on ei­ther idea be­cause I keep flip­ping back and forth about which seems most im­por­tant.”

Mike: “Can you try calcu­lat­ing the time re­quired for the learn­ing, and the time saved af­ter­ward, to calcu­late which is bet­ter?”

Lyra: “So I tried do­ing that ear­lier. I es­ti­mated the ar­chi­tec­ture learn­ing would take me fifty to a hun­dred hours to do, and save maybe twenty or thirty hours a year. On the other hand, one of the smaller pro­jects would only take five or ten hours, and it would save me a few min­utes a day, which adds up to ten or fif­teen hours a year.”

Mike: “That sounds like the smaller pro­ject is clearly a bet­ter deal—it would pay for it­self within one year, while the big­ger pro­ject would need more like three years to break even.”

Lyra: “I know, but the big­ger pro­ject feels im­por­tant any­way. I think… the big­ger pro­ject isn’t just about sav­ing time. I care about it be­cause it also opens up the op­tion to do new things that I can’t do right now. But I’m not sure if that benefit is big enough to make it worth do­ing.”

Mike: “Could you run an ex­per­i­ment for five hours or so to see if it seems like you’re able to do new things?”

Lyra: “Yeah, that sounds good. If it looks like I’m not, then I can go ahead with the smaller pro­ject since that’s bet­ter for sav­ing time.”

Even though Lyra didn’t fol­low her num­bers ex­actly, they were helpful for clar­ify­ing the de­ci­sion.

3. Job hunting

Mark has just grad­u­ated from uni­ver­sity, and he wants to com­plete an ML mas­ters to be an en­g­ineer at an AI safety org or earn to give if that doesn’t work out.

How­ever, the pro­grams cost be­tween $15,000 and $24,000, and Mark isn’t com­fortable go­ing into debt. So his plan is to get a job now, ap­ply for the mas­ters’ pro­grams in the fall, and save up money un­til he starts the fol­low­ing year.

Mark’s pre­vi­ous sum­mer job has offered him a full-time role for $15 an hour. How­ever, his un­der­grad the­sis su­per­vi­sor en­courages him to look into soft­ware en­g­ineer­ing jobs, which the su­per­vi­sor thinks he’s qual­ified for. Un­for­tu­nately, he doesn’t have enough time to take the sum­mer job while also ap­ply­ing.

Mark isn’t sure about it, but his su­per­vi­sor con­vinces him to look into some job post­ings for lo­cal po­si­tions. So Mark puts to­gether the fol­low­ing es­ti­mates.

Based on those num­bers, he de­cides it’s worth de­lay­ing start­ing at his pre­vi­ous job to spend a cou­ple of months ap­ply­ing to en­g­ineer­ing roles. If it works out, he’ll earn a lot more. If the job hunt doesn’t seem promis­ing af­ter a few months, he can go back to the other role.

Bal­l­park Estimates

In each of these ex­am­ples, putting num­bers on the un­cer­tainty helped the per­son pri­ori­tize. Es­ti­mat­ing didn’t cleanly de­cide their pri­ori­ties, but it re­duced un­cer­tainty. It re­vealed blind spots, such as miss­ing im­por­tant con­sid­er­a­tions or think­ing two things were com­pa­rable when one was ac­tu­ally way more im­por­tant. By quan­tify­ing, rank­ing, or crunch­ing num­bers, they were able to make a bet­ter guess at what should be pri­ori­tized.

Quan­tify­ing his ex­pected value and time re­quired helped Will in­crease his ex­pected value per hour of work 2x com­pared to work­ing on one of the pa­pers at ran­dom. Quan­tify­ing re­turn on in­vest­ment for her time learn­ing helped Lyra iden­tify what fac­tors she was over­look­ing, so now she can eval­u­ate whether the pro­ject is worth­while. Mark de­cided to ap­ply to more am­bi­tious jobs based on his num­bers, and got a role mak­ing >50% more af­ter three months of ap­ply­ing.

Clas­si­cally, Fermi es­ti­mates are back-of-the-en­velope calcu­la­tions in­tended to quickly ap­prox­i­mate the cor­rect an­swer, usu­ally when the real an­swer is difficult to get. Here, peo­ple es­ti­mated val­ues and costs of differ­ent op­tions, so they could ap­prox­i­mate the re­turn on their in­vest­ment and com­pare op­por­tu­nity costs.

Since we’re fre­quently pri­ori­tiz­ing amid un­cer­tainty, even mod­er­ately re­duc­ing that un­cer­tainty im­proves de­ci­sions. Often you have some data eas­ily available even when you feel un­cer­tain. Some­times this is just mak­ing your in­tu­ition con­crete. Some­times it is ac­tu­ally gath­er­ing data and crunch­ing num­bers. You should take care not to be over­con­fi­dent in your es­ti­mates, but even to­tally made up num­bers can some­times be use­ful, such as when you want to make a de­ci­sion be­tween com­pet­ing in­tu­itions.

So, what re­turns do you get on the time and effort in­vested? How does this com­pare with your other op­tions?

V. 8020 Rule

1. Work­ing hours experiment

Bill was frus­trated by a con­sis­tent dip in en­ergy each af­ter­noon. He felt less mo­ti­vated dur­ing this time, and slug­gish and slow even when he forced him­self to work.

Work­ing out in the af­ter­noon helped him feel more en­er­getic af­ter­ward, but tak­ing a forty-five-minute break made him feel like he needed to work late.

He knew that sub­jec­tive ex­pe­rience doesn’t always match ac­tual out­put, so he tried quickly record­ing how many words per hour he wrote for a week. Bill also noted how he sub­jec­tively felt dur­ing that time. At the end, the data sug­gested that his out­put dropped by nearly half dur­ing that pe­riod, and only grad­u­ally picked back up over the later af­ter­noon.

He kept record­ing the data while he took some work­out breaks. Although the data was noisy, he found that he got about as much done on days when he worked out and days when he didn’t.

Given that, he de­cided to work out each af­ter­noon with­out feel­ing obli­gated to work late.

2. Op­ti­miz­ing school

Sarah is a col­lege fresh­man ask­ing Alice, a se­nior who works in the same lab, for tips on how to suc­ceed in col­lege.

Sarah: “So, what are the most im­por­tant things you do to get good grades?”

Alice: “Umm, I plan each day the night be­fore so I know ex­actly what I need to do, and then I set aside a cou­ple of hours when I turn off my phone and study with­out any dis­trac­tions. That’s big. I usu­ally do the most im­por­tant task first so that I don’t risk run­ning out of time. Oh, and I have a ques­tion in mind while I’m do­ing re­search, so that I don’t lose too much time go­ing down rab­bit holes.”

Sarah: “Is there some­thing else that helps you re­li­ably man­age your work­load?”

Alice: “So, I start my plan­ning by look­ing at which pro­jects are worth a lot of a grade or that I care a lot about learn­ing, and I choose which pro­jects de­serve the most time and which to just do the bare min­i­mum. For ex­am­ple, go­ing from an okay pa­per to a great one takes a lot more work, so I’ll only do that if I re­ally care about the pa­per. Other­wise, I’ll wait to start the pa­per un­til the day be­fore it’s due, and then race like crazy. It forces me to get the pa­per done with­out spend­ing too much time on it.”

Sarah: “Thanks, Alice. That sounds helpful. I’ve been feel­ing re­ally over­whelmed by tak­ing three re­ally hard classes, and I re­ally want to get As in all of them. I’m a bit of a perfec­tion­ist, I know.”

Alice: “It’s great that you’re ask­ing for ad­vice, Sarah, that will prob­a­bly help you figure col­lege out way faster than I did. But I want to ask, why do you care about get­ting good grades?”

Sarah: “Wha-what?”

Alice: “Why are As the thing you’re aiming for right now?”

Sarah: “That doesn’t even make sense. Grades are just what we’re sup­posed to do here. Wait, I’ll try to work out the rea­son… Be­cause in school, grades are how we know we’re do­ing well or where we need to im­prove. And how fu­ture em­ploy­ers or grad schools know that we’re good.”

Alice: “That’s fair, but it’s only a small part of what peo­ple will care about in the fu­ture. You’re only a fresh­man, but you’re already work­ing in this lab and your re­search seems re­ally promis­ing. You ob­vi­ously love do­ing it. But you’re only do­ing five hours a week here be­cause you say you don’t have enough time to do this and study. If you in­stead spent a lot more time on re­search and did re­ally cool things there, I bet both grad schools and fu­ture em­ploy­ers would care about that more than a 4 point GPA.”

Sarah: “Hmm.”

Alice: “I men­tioned ear­lier how im­por­tant it is to de­cide which parts of a pro­ject de­serve more time, and which to just put in the min­i­mum. Well, it’s even more im­por­tant to care­fully choose what pro­jects or classes are worth a lot of time, and when it would be bet­ter to do the min­i­mum you can in some classes, so that you can in­vest heav­ily in oth­ers.”

Sarah: “I need to think about this. What you’re say­ing kind of makes sense, but I’m wor­ried that if I’m do­ing the bare min­i­mum, my grades will drop too much.”

Alice: “Good things to con­sider. I’m ap­ply­ing to med school, so my grades mat­ter. But I chose to take easy classes for my gen eds and elec­tives, so that I can put in a lot of time here at the lab with­out dam­ag­ing my GPA. That’s the right de­ci­sion for me. I’ll bet it’s worth­while for you to spend some time think­ing about what you want to be perfec­tion­ist about.”

3. High-value rest time

In my in­ter­view with Tara Mac Auley, she ad­vised try­ing a bunch of leisure ac­tivi­ties to de­cide which are most valuable for you.

“If you take time to rest and you come back, and it doesn’t feel bet­ter then prob­a­bly the ways that you’ve cho­sen to rest aren’t in fact the most restora­tive things you could be do­ing. And so I would sug­gest try­ing a lot of differ­ent things: a lot of differ­ent types of so­cial ac­tivi­ties, or phys­i­cal ac­tivi­ties, or in­tel­lec­tu­ally en­gag­ing ac­tivi­ties.
I did this a lot when I was in my early 20s. I picked a ran­dom event from ev­ery day for about two months, and I just had to go to whichever thing came up. And then I would write down be­fore­hand whether I thought I would en­joy it and feel drained or re­freshed from that ac­tivity. And then I would com­pare af­ter­ward what I ac­tu­ally felt and, I don’t know, that was re­ally in­for­ma­tive and good for me.”

She used this type of pro­cess to iden­tify the ac­tivi­ties that best leave her re­ju­ve­nated and rested.

“Be­ing near wa­ter and swim­ming, but not in a swim­ming pool, it has to be nat­u­ral wa­ter. Be­ing in na­ture. Read­ing a book, es­pe­cially read­ing a book in a park or by a lake or some­thing like that. Spend­ing qual­ity time with close friends or fam­ily just hav­ing a con­ver­sa­tion for an hour with no par­tic­u­lar goals, I find re­ally re­ju­ve­nat­ing. And eat­ing a re­ally nice meal; one where I can kind of sa­vor all of the differ­ent tastes and tex­tures….I go out danc­ing a lot on my own, to go and see mu­sic artists that I en­joy, and I just dance like a crazy per­son un­til I’m re­ally tired and then I go home, and that’s amaz­ing.”

8020 Rule

In each of these ex­am­ples, the per­son wanted to pri­ori­tize the most valuable sub­set of pos­si­ble ac­tions. By iden­ti­fy­ing the higher-value ac­tions, they could get more done for their effort.

Bill did an ex­per­i­ment to find out which hours of work pro­vided the most value, so he could make bet­ter de­ci­sions about when to work. Sarah pri­ori­tized the high­est value work to get good grades, and started think­ing about how much more valuable that effort would be if she pri­ori­tized the high­est value goals to be­gin with. Tara ex­per­i­mented to find out which ac­tivi­ties were the high­est value fun for her, which she can now ex­ploit 8020 style.

Based on the idea that 80% of an out­put comes from 20% of the in­put, the 8020 rule sug­gests that the value per unit of effort varies a lot across differ­ent ac­tions. Be­cause out­puts vary so much, ex­plore more can un­earth dra­mat­i­cally bet­ter op­tions. So, similar to Tara, you need to try many ac­tions first in or­der to effec­tively iden­tify the top-perform­ing sub­set. Once you’ve iden­ti­fied which ac­tions are most valuable, you can nar­row your fo­cus to just that sub­set. Then your out­put will in­crease sig­nifi­cantly for the same effort.

So, what gives you the most value for the least effort? What can you cut with min­i­mum loss so that you have ex­tra re­sources to put to­ward what’s most valuable?


All of these sto­ries are of how peo­ple tried to iden­tify the most valuable ac­tions they could take to ac­com­plish their goals. They re­duced un­cer­tainty, said no to other ac­tions, and made choices based on their best guesses.

You might be tempted to say these ex­am­ples don’t feel im­por­tant. That choos­ing which skill to learn or over­com­ing anx­iety can’t change the world. And maybe you’re right, if you only look at that one step.

If you put all of these to­gether, you have a mind­set that searches for the most valuable goals, builds mod­els to effec­tively ac­com­plish them, iter­a­tively tests as­sump­tions against the world, log­i­cally weighs the op­por­tu­nity cost, and ju­di­ciously spends time and effort to get the most im­pact pos­si­ble. That mind­set touches all your de­ci­sions.

And that’s pri­ori­ti­za­tion.

Be­cause pri­ori­ti­za­tion isn’t some­thing you do once a month. It’s not a mag­i­cal abil­ity that lets you do ev­ery­thing—quite the op­po­site, in fact. It’s the gut-deep sense that your time and effort are limited and you need to choose what to do, be­cause you can’t do ev­ery­thing.

But when you do that? When you put all of your rea­son and tools to the task of choos­ing the most valuable goals?

Then we have a chance. Choose im­por­tant goals, and you could save lives from dy­ing of malaria or build a fu­ture where pan­demics don’t wipe out hun­dreds of thou­sands. Ac­com­plish those goals, and the world be­comes bet­ter. If you need to take care of your own men­tal health or build skills first, then do it. You’re still nudg­ing the world in the right di­rec­tion.

And if you don’t? …Then we’re still right where we are now. We’ve lost out on some of the good­ness and won­der the world could have had. There’s the sense of be­ing so close and just miss­ing what could have been.

That’s why I want to con­vey the mind­set of what it feels like to pri­ori­tize.

So, here are five ques­tions to take with you. Use them to make the world bet­ter.

  1. What steps will ac­tu­ally make you likely to achieve your goals?

  2. What is the first quick test you could cre­ate to get feed­back and iter­ate?

  3. What are the one or two things you could change about your­self or your en­vi­ron­ment to ac­com­plish twice as much?

  4. What re­turns do you get on the time and effort in­vested? How does this com­pare with your other op­tions?

  5. What gives you the most value for the least effort? What can you cut with min­i­mum loss so that you have ex­tra re­sources to put to­ward what’s most valuable?

En­joyed the piece? Sub­scribe to EA Coach­ing’s newslet­ter to get more posts de­liv­ered to you.