Hammertime Day 10: Murphyjitsu

Link post

This is part 10 of 30 in the Ham­mer­time Se­quence. Click here for the in­tro.

Like, so pes­simistic that re­al­ity ac­tu­ally comes out bet­ter than you ex­pected around as of­ten and as much as it comes out worse. It’s ac­tu­ally re­ally hard to be so pes­simistic that you stand a de­cent chance of un­der­shoot­ing real life.

Later in the day I will put up an open thread about the first cy­cle of Ham­mer­time.

We finish up the first cy­cle with an­other post on plan­ning. Mur­phyjitsu is CFAR’s method for plan­ning which asks us to try to be so pes­simistic as to un­der­shoot real life.

Day 10: Murphyjitsu

Mur­phy’s Law states that any­thing that can go wrong will go wrong.

For our Man­darin-speak­ing read­ers, here’s a use­ful mnemonic: Mur­phy transliter­ates as 墨菲 (mo fei), which is ho­mophonous to 莫非, “what if?” That’s why I think of Mur­phy’s Law as the What If Law.

In the course of mak­ing plans, Mur­phyjitsu is the prac­tice of strength­en­ing plans by re­peat­edly en­vi­sion­ing and defend­ing against failure modes un­til you would be shocked to see it fail. Here’s the ba­sic setup of Mur­phyjitsu:

  1. Make a plan.

  2. Imag­ine that you’ve passed the dead­line and find out that the plan failed.

  3. If you’re shocked in this sce­nario, you’re done.

  4. Other­wise, simu­late the most likely failure mode, defend against it, and re­peat.

The first im­por­tant sub-skill of Mur­phyjitsu is In­ner Sim – the abil­ity for Sys­tem 1 to simu­late failure modes.

In­ner Sim

I have the sus­pi­cion that ev­ery­one is se­cretly a mas­ter at In­ner Sim, the abil­ity to in­stantly simu­late failure. Imag­ine a friend de­clares to you their New Year’s Re­s­olu­tion: to write a novel, to go on a keto diet, to write a month-long se­quence on in­stru­men­tal ra­tio­nal­ity.

Now, listen for that in­ter­nal scoffing – your Sys­tem 1 in­stantly pro­lifer­ates the fu­ture with all man­ner of ob­sta­cles. That’s In­ner Sim at work.

If you’re any­thing like me, In­ner Sim is bet­ter at pre­dict­ing other peo­ple’s failure modes than your own. The men­tal move that helps ap­ply In­ner Sim in­tro­spec­tively is es­sen­tially Out­side View: take your plan and imag­ine an­other per­son made them. What will go wrong?

Welp Mentality

In­ner Sim does sur­pris­ingly lit­tle on its own.

I had a con­ver­sa­tion with a ra­tio­nal­ist friend (let’s call him “Alex”) that went some­thing like this:

Alex: What’s both­er­ing you?

Me: I’ve been ter­ribly un­pro­duc­tive. I’m pro­cras­ti­nat­ing on fel­low­ship es­says … they’re due in two weeks, and ev­ery time I think about math these es­says pop up in my head.

Alex: Why?

Me: I es­sen­tially finished them, but I still have to edit it. Copy-edit­ing is so te­dious, and ev­ery run I make through my writ­ing, it looks even more awk­ward than it did the pre­vi­ous time.

Alex: What do you pre­dict will hap­pen?

Me: Well … I’m go­ing to put the es­says off un­til two days be­fore the dead­line, edit for ten min­utes when I start feel­ing the pres­sure, and sub­mit them. Un­til then, I won’t get any re­search done.

Alex: So…?

Me (shrugs): Sucks, right?

Alex breaks down in laugh­ter.

I call this Welp Men­tal­ity. Welp Men­tal­ity is notic­ing that your plans are likely to fail catas­troph­i­cally, or run over­time, or take 10x as much effort as you thought, and then shrug­ging non­com­mit­tally. Welp.

Welp Men­tal­ity is know­ing and ac­cept­ing as a fact of life that ev­ery build will re­lease two months late. That you’ll end up half-ass­ing prob­lem sets and es­says start­ing mid­night be­fore the dead­line. That your cur­rent ex­er­cise plan will prob­a­bly pe­ter out. I had an old motto for Welp Men­tal­ity: “Due to­mor­row? Do to­mor­row.”

Murphyjitsu

Mur­phyjitsu is the as­tound­ing no­tion that if you can pre­dict a failure mode, you can do some­thing about it!

If your builds re­lease two months late ev­ery time, you can move the re­lease date, or cut fea­tures, or hire more en­g­ineers. If you know you’re only go­ing to spend six hours on a prob­lem set the night be­fore the due date, at very least you might as well just set a six-hour Yoda Timer for it, do it at a con­ve­nient time, and sub­mit what­ever you end up with.

In my fel­low­ship es­say case, I de­cided to spend ten min­utes edit­ing and sub­mit the thing im­me­di­ately. The re­lief of get­ting two weeks of my life back was pal­pable.

Pick a plan you have for the near fu­ture. Mur­phyjitsu it. Pull out all the stops: Ar­range so­cial pres­sure to keep you on track. Dou­ble the amount of time you spend. Set cal­en­dar and phone re­minders. Mur­phyjitsu only stops when you would be shocked if the plan fails.

Daily Challenge

Mur­phyjitsu a cen­tral life goal. Are there glar­ing failure modes you haven’t defended against?