# sky’s Shortform

• Quick thoughts on turn­ing per­centages back into people

Oc­ca­sion­ally, I ex­per­i­ment with differ­ent ways to grok prob­a­bil­ities and statis­tics for my­self, start­ing from the ba­sics. It also in­volves pay­ing at­ten­tion to my emo­tions, and imag­in­ing how differ­ent ex­pla­na­tions would work for differ­ent stu­dents. (I’m of­ten a men­tor/​work­shop pre­sen­ter for col­lege stu­dents). If your brain is like mine or you like see­ing how other peo­ple’s brains work, this may be of in­ter­est.

One trick that has worked well for me is turn­ing %s back into people

Ex­am­ple: I think my Pro­ject X can solve a prob­lem for more peo­ple than it’s cur­rently do­ing. I have a sur­vey (N=1200) which says I’m cur­rently solv­ing a prob­lem for 1% of the peo­ple im­pacted by Is­sue X. I think I can definitely make that num­ber go up. Also, I re­ally want that num­ber to go up; 1% seems so paltry.

I might start with:Ok, how likely do I think it is that 1% could go up to 5%, 10%, 20%?

But I think this is the wrong ques­tion to start with for me. I want to in­form my in­tu­itions about what is likely or prob­a­ble, but this all feels su­per hy­po­thet­i­cal. I know I’m go­ing to want to say 20%, be­cause I have a bunch of ideas and 20% is still low! The %s here feel too fuzzy to ground me in re­al­ity.

Alter­na­tive: Turn 1% of 1200 back into 12 peo­ple

This is 12 peo­ple who say they are pos­i­tively im­pacted by Pro­ject X.

This helps me re­mem­ber that no one is a statis­tic. (A post which may have in­spired this idea to be­gin with). So, yay, 12 peo­ple!

But go­ing from 1% to 5% still sounds un­am­bi­tious and un­satis­fy­ing. I like am­bi­tious, tena­cious, hope­ful goals when it comes to peo­ple get­ting the solu­tions they’re look­ing for. That’s the whole point of the pro­ject, af­ter all. Some­times, I can phys­i­cally feel the stress over this ten­sion. I want this num­ber to be 100%! I want the prob­lem solved-solved, not kinda-solved.

At this point, maybe I could re­mind my­self or a stu­dent that “should­ing at the uni­verse” is a recipe for frus­tra­tion. I love that con­cept, and some­times it works. But of­ten, that’s just an­other way of should­ing at my­self. The fact re­mains that I don’t want to be less am­bi­tious about solv­ing prob­lems that I know are real prob­lems for real peo­ple.

I try the per­cents-to-peo­ple tech­nique again:

• Turn 5% of 1200 back into 60 peo­ple. Oh. That’s 48 ad­di­tional peo­ple. Also no­tice: it’s only 60 peo­ple if we’re talk­ing about 48 ad­di­tional peo­ple, while los­ing 0.

• Turn 10% back into 120 peo­ple. 108 ad­di­tional peo­ple, while los­ing 0.

• Turn 20% back into 240 peo­ple. 228 ad­di­tional peo­ple, while los­ing 0.

• So, an in­crease of 5% or 20% is the differ­ence be­tween 48 or 228 ad­di­tional peo­ple reached. I know about this pro­gram be­cause I work on it, and I know how much goes into Pro­ject X right now to reach 12 peo­ple. I’m sure there are things we could do differ­ently, but are they differ­ent enough to reach 228+ ad­di­tional peo­ple?

Now this feels differ­ent. It’s hum­bling. But it piques my cu­ri­os­ity again in­stead of my frus­tra­tion: how would we at­tempt that? Could we?

• What else do I need to know, to figure out if 60 or 120 or 240 (...or 1000, or 10000) is any­where within the realm of pos­si­bil­ities for me?

• Do I have a clear idea about what my bot­tle­necks or mis­takes are in the sta­tus quo, such that I think there are 48 more peo­ple to reach (while still reach­ing the 12)? What pro­cesses would need to change, and how much?

• This im­me­di­ately brings up the re­sponse, “That de­pends on how long I have.” (Woot, now I’ve just grokked why it’s use­ful to time-bound ex­am­ples for com­par­i­son’s sake). We could call it 1 year, or 3, or 10, etc. I per­son­ally think 1-3 years is usu­ally eas­ier to con­cep­tu­al­ize and op­er­a­tional­ize.

• Also, what­ever I do next, it’s ob­vi­ously go­ing to take no­table effort. I know I can only do so much work in a day. (I prob­a­bly hate this truth the most. This is definitely where I re­mind my­self not to should at the uni­verse). Now I won­der, is this definitely the pro­gram where I want to fo­cus my effort for a while? Why? What if there are prob­lems up­stream of this one that I could put my effort to­ward in­stead? …aha, now my un­der­stand­ing of why peo­ple care about cause pri­ori­ti­za­tion just got deeper and more per­son­ally in­tu­itive. This is a topic for an­other post.

To re­turn to per­centages, here’s one more ex­am­ple. Per­centages can also feel daunt­ing in­stead of un­am­bi­tious:

• Go­ing from 12 to 60 peo­ple is a 400% in­crease. (Right? I haven’t mis­calcu­lated some­thing ba­sic? Yes, that’s right; thank you, on­line calcu­la­tors). 400%! Is that mad­ness?

• Turn ’400% in­crease’ back into 4 ad­di­tional peo­ple reached, for ev­ery 1 per­son reached now.

That may still be daunt­ing. But it may be eas­ier to make es­ti­mates or com­pare my in­tu­itions about differ­ent ac­tion plans this way.

If you (or your stu­dents) are like me, this is a use­ful ap­proach. It gets me into the headspace of imag­in­ing cre­ative pos­si­bil­ities to solve prob­lem X, while still ground­ing my­self within some con­crete pa­ram­e­ters rather than los­ing my­self to should­ing.

• We­bi­nar to­mor­row: ex­plor­ing solu­tions jour­nal­ism [for EA writ­ers]:

If EA jour­nal­ists and writ­ers are plan­ning to cover EA top­ics, I think a solu­tions jour­nal­ism an­gle will usu­ally be the most nat­u­ral fit.

The Solu­tions Jour­nal­ism Net­work “train[s] and con­nect[s] jour­nal­ists to cover what’s miss­ing in to­day’s news: how peo­ple are re­spond­ing to prob­lems.”

The Solu­tions Jour­nal­ism Net­work is hav­ing a we­bi­nar to­mor­row: https://​​zoom.us/​​we­bi­nar/​​reg­ister/​​WN_Qcbxqd-uRvyvy1OnvVaIPg

Solu­tions journalism

“Can be char­ac­ter-driven, but fo­cuses in-depth on a re­sponse to a prob­lem and how the re­sponse works in mean­ingful detail

• Fo­cuses on effec­tive­ness, not good in­ten­tions, pre­sent­ing available ev­i­dence of results

• Dis­cusses the limi­ta­tions of the approach

• Seeks to provide in­sight that oth­ers can use”

This is still a less com­mon me­dia an­gle. The qual­ity of cov­er­age will clearly still vary a lot de­pend­ing on one’s re­search, ed­i­to­rial in­put, etc, but this is a bet­ter fit than many other me­dia an­gles one could take to cover top­ics of in­ter­est to you in EA.