An On­to­logy of Sys­temic Fail­ures: Dragons, Bull­shit Moun­tain, and the Cloud of Doom

Core Claim

I as­sert that a lot of value can be achieved by cat­egor­iz­ing sys­temic fail­ures into four broad cat­egor­ies.

For the sake of pith­i­ness, I will name them “bugs”, “dragons”, “bull­shit moun­tain”, and “the cloud of doom”.

A Bug is the simplest kind of fail­ure: you have a single cause, and a single symp­tom. Fix­ing a bug is pretty easy, com­pared to other modes—you just fix the cause, and the symp­tom goes away. Bugs aren’t even really sys­temic fail­ures, since they don’t in­volve any cross-talk between mul­tiple causes or ef­fects, but they are in­cluded here for com­plete­ness.

A Dragon is like a Bug, ex­cept that in­stead of a single symp­tom, there are mul­tiple seem­ingly in­de­pend­ent symp­toms. Be­cause the symp­toms seem large and di­verse, a Dragon will of­ten seem far more daunt­ing than it ac­tu­ally is—al­though most Dragons are still pretty large and sig­ni­fic­ant causes. Deal­ing with a Dragon simply in­volves some par­tic­u­larly Heroic-type identi­fy­ing that there’s a Dragon, hunt­ing it down, and slay­ing it.

Bull­shit Moun­tain is ba­sic­ally the op­pos­ite of the Dragon. There is one huge, un­bear­able, pain­ful symp­tom, that every­one knows about and every­one wishes would just GO AWAY. But no one can get any trac­tion on it. This is be­cause that symp­tom is ac­tu­ally be­ing con­trib­uted to by a thou­sand little causes, all of which only con­trib­ute a little—so mak­ing pro­gress on any one of them feels like it doesn’t help much, if at all. The only way to solve Bulshit Moun­tain is for every­one in the org to roll up their sleeves, get a shovel, pick some little corner of Bull­shit Moun­tain to work on, and start shov­el­ing—and not stop un­til the prob­lem gets no­tice­ably bet­ter, even if their work doesn’t seem to be do­ing much to con­trib­ute to the im­prove­ment.

Fin­ally, we have the Cloud of Doom. The Cloud of Doom is where you have a thou­sand tiny causes, each of which mean­ing­fully con­trib­ute to each of a thou­sand little symp­toms, which to­gether make the whole sys­tem feel un­work­able. The only way to fix the Cloud of Doom is to pour an un­godly amount of Slack into the sys­tem, and hope the cloud shakes loose and blows away—oth­er­wise, every­one needs to just throw up their hands and aban­don the whole mess.

A good ques­tion to ask your­self, when try­ing to tackle a seem­ingly in­sur­mount­able mess of prob­lems, is: is this a Dragon, Bull­shit Moun­tain, or a Cloud of Doom?

Slay­ing Dragons

If the prob­lem is a Dragon, are you the one qual­i­fied to slay it? If you are, what fur­ther in­form­a­tion or re­sources do you need? Do you need a team? Are you the right per­son to lead that team?

Be­cause Dragons are single-cause prob­lems, they re­spond well to a single per­son and a single plan. Most or­gan­iz­a­tions hope (and there­fore pre­tend) that most of their prob­lems are Dragons, and most or­gan­iz­a­tional prob­lem-solv­ing is ded­ic­ated to find­ing Dragons (or mak­ing prob­lems look like Dragons) and then pay­ing a few core Her­oes big bucks to slay them.

Basic­ally, slay­ing Dragons is a Solved Prob­lem in the or­gan­iz­a­tional world; most ap­par­ent lack of suc­cess in­volves mis-identi­fy­ing Bull­shit Moun­tains and Clouds of Doom as Dragons, in the na­ive hope that they’ll turn out to be Dragons any­ways and there­fore can be solved by a pro­cess that the sys­tem knows how to im­ple­ment.

Shov­el­ing Bull­shit Mountain

If your prob­lem is Bull­shit Moun­tain, do you have enough buy-in to get enough people to start shov­el­ing? Are you the right cheer­leader to keep people mo­tiv­ated? Does the team still care enough to even want to solve the prob­lem? These are way harder prob­lems to tackle than the Dragon prob­lems lis­ted above, so ex­pect a lot of people to hand­wave and con­vince you that the prob­lem is a Dragon (if they want to sig­nal buy-in for solv­ing it) or a Cloud of Doom (if they don’t).

Deal­ing with Bull­shit Moun­tain as if it was a series of Dragons is what causes “death marches” in the tech in­dustry. It’s what makes trans­form­at­ive “busi­ness cul­ture” ex­per­i­ments seem to work tem­por­ar­ily (by clear­ing out the moun­tain and re­pla­cing it with a new sys­tem, which then be­gins ac­cu­mu­lat­ing its own Bull­shit). It’s the biggest con­trib­utor to low em­ployee mor­ale that a mid- to large-size “healthy” or­gan­iz­a­tion can have. (In fact, the trans­ition from Bull­shit Moun­tain into a Cloud of Doom is prob­ably the tip­ping point for an or­gan­iz­a­tion be­com­ing un­sal­vage­able.)

Sur­viv­ing The Cloud of Doom

Fin­ally, we have the Cloud of Doom. Every or­gan­iz­a­tion has one. Some are big, some are small, some are more toxic than oth­ers. But an or­gan­iz­a­tion with 0% of its prob­lems in a Cloud of Doom is an or­gan­iz­a­tion that has not yet had to do any­thing ac­tu­ally real.

So, you live with it. And the or­gan­iz­a­tion be­gins to de­velop other prob­lems—mostly bugs, but some Dragons and a few Bull­shit Moun­tains here and there. And as the Dragons get big­ger and lay waste to more coun­tryside, and as the Bull­shit Moun­tains tower higher and higher over­head, they start to in­ter­weave and cor­res­pond, feed­ing the Cloud of Doom.

Even­tu­ally, the Cloud of Doom be­gins to ac­tu­ally choke the life out of your or­gan­iz­a­tion. That’s when you have a single choice: in­ject lots and lots of Slack, or leave.

In­ject­ing lots and lots of Slack ba­sic­ally means “do­ing less with more”, which al­most no one be­lieves is the cor­rect choice. But if you aren’t burn­ing everything down and start­ing over some­where else, it’s the only choice. If you can’t live with your Cloud of Doom, and you won’t flee it, you’re go­ing to have to stop feed­ing it and let it blow away.

Har­ness­ing the Cloud of Doom

One thing you CAN do, if you’re par­tic­u­larly vi­cious, is con­vince people that your Cloud of Doom is ac­tu­ally just Bull­shit Moun­tain, and use that to ex­tract work from your sub­or­din­ates. You need to be ex­cep­tion­ally clever (in a Raoian so­ciopathic sense) in or­der to pull this off, be­cause you ba­sic­ally need to man­age both sides of the in­form­a­tion and ef­fort flow: you have to keep every­one be­liev­ing that there’s a Bull­shit Moun­tain that they’re bit­ing into, AND you have to re-dir­ect and man­age their ac­tual ef­forts so that they be­ne­fit your cov­ert goals. Note that this is doable whether the ac­tual prob­lem is in fact a Cloud of Doom or merely Bull­shit Moun­tain, and that ap­ply­ing this very pro­cess to Bull­shit Moun­tain is one of the more com­mon things that turns Bull­shit Moun­tain into a Cloud of Doom in the first place.

If I catch you do­ing this within any or­gan­iz­a­tion that I am aligned with, I con­sider it within my rights to des­troy you.

Dragonslay­ers are Gry­ffindor, Shov­el­ers are Hufflepuff

No­ti­cing what kind of prob­lem you have the right tem­per­ment to solve is key to avoid­ing burn-out. Dragonslay­ing is glam­or­ous, high-praise work; shov­el­ing Bull­shit Moun­tain is thank­less and gruel­ing, and the per­son who fi­nally gets the praise is usu­ally the guy that did the least ac­tual work. From my per­spect­ive, Pro­ject Huffle­puff was in many ways a dir­ect at­tempt to train up people who could handle the Ra­tion­al­ist-and-EA-com­munity’s Bull­shit Moun­tain be­fore it turned into a Cloud of Doom. (Whether it’s too late now or not is a mat­ter for the Slyther­ins to con­vince you; I will say no more on this today.)

One big prob­lem with shov­el­ers is that they still ex­pect praise and re­ward for shov­el­ing Bull­shit Moun­tain; no one seems to be telling them that it ain’t gonna hap­pen. (The smart ones fig­ure this out on their own, and either go away or grit their teeth and get to work any­ways.) As a sys­tem for mo­tiv­at­ing people to shovel, any­thing like Pro­ject Huffle­puff is doomed to fail from the be­gin­ning. What you need is a way to identify the people who are already do­ing the work, and make sure they stay fun­ded and sup­por­ted and nur­tured. (Whether this prob­lem it­self is a Bull­shit Moun­tain or a Cloud of Doom is, again, left as an ex­er­cise for the reader. A Dragon it ain’t, or Pro­ject Huffle­puff would have worked.)


So, any­way, yeah. Here we are. Do what you will with it. Or don’t; I’m not your dad.

[Epistemic status]