Haven’t thought about my piece cited here (the GUTs thing from 2018, contraptions.venkateshr…) in a long time.

I think what Ben Reinhardt is calling “fat ideas” here should really be relabeled thorough ideas (cf: ETTO, efficiency thoroughness tradeoff). Ideas that require that you explore (in a directed way) a large frontier rather systematically, and cover a lot more bases without cutting corners in order to get it right, because do-overs are not cheap. “Lean” can afford to sacrifice thoroughness because do-overs are cheap. It gets conflated with efficiency. Lean is typically not thorough, but it’s not necessarily efficient either. “Lean” imo is an evocative but bad term to apply to early stages. It really only applies to mature stage ops that are already orbiting late-stage local-optimum process attractors.

The whole “lean startup” model was rife with fallacies and deep conceptual problems. It was “efficient search in cheap-do-over space fueled by cheap capital.” The apparent rhyming with toyota lean, lean six sigma, OODA etc. was bullshit. Even in mature industry, DMAIC (the process optimization part of lean six sigma) kinda worked, but DMEDI (the “innovation” part) never did. It was pure bullshit innovation theater dreamt up by consultants trying to upsell (DMAIC was slightly less bullshit process optimization almost-theater — overpackaged and gamified DoE basically — design of experiments).

Lean-fat as currently used should be replaced by efficiency-thoroughness/ETTO. Lean should be reserved for late stage DMAIC-like cost-efficiency drives. Fat should be reserved for undirected exploratory potential. If I had more energy, I’d be holy-warrior evangelizing this message. A good career-maker for the right kind of consultant just starting out.

For my bit role in this discourse, I have a small but critical update to how I define “fat” relative to the GUTS model these days — it’s not just stocks of surpluses all along a (directed) value chainfrontierfan, but high disorganization of those surpluses. Both efficiency and thoroughness are downstream of fatness in this sense of disorganization. “Thoroughness” unlike fatness proper has a directed organization to it A compass indicating a true north and inducing how resources are arrayed.

High-capex hard-tech things look pseudo-fat because they have to organize a lot of resources and kinda run waterfall without end-to-end feedback for extended periods (rocket launches are not cheap even at Elon levels of willingness to blow things up frequently). The resources marshaled should be considered staged on a notional Gantt chart rather than stored in an unallocated way as fat.

Orderliness comes from direction. True fat is generally almost directionless (at best it has some anisotropic grain to it), but high potential. It has low entropy at the level of “energy density” relative to the environment, but high entropy at higher levels of organization.

Like a poorly organized kitchen has a lot of calories that can be converted into edible meals, but it isn’t set up to do anything well immediately. You have to first pick a cooking project before you can organize rationally.

Depending on what you pick, you may need to organize “efficiently” or “thoroughly” and do more or less shopping to fill the gaps in the mise en place. Some dishes take a lot of time and setup in prep (ie thoroughness) before you can begin cooking. Other dishes you can just get going with little to no prep (efficient), and make adjustments along the way, including even rushing to the store last minute to get something.

The kitchen might have a “grain” to it making it better pre-staged for some things more than others (my kitchen is well set up for Indian, Thai, Italian cooking for eg, and poorly set up for Mexican and Korean, and not at all for French). The “grain” is some minimal directionality fat can have. So efficiency/thoroughness projects are defined relative to the grain of the fat to some degree. What’s efficiently doable given one grain may require more thorough setup wrt another grain.

This “grain” of fat is really the protocolization of the environment, which is one big idea that’s occupying a lot of my bandwidth at the moment. Protocolization is fat-scaffolding. Just enough structure to create a somewhat opinionated grain, but not enough to prevent total disorganization due to directionlessness of anticipated usage. So traffic systems for example are protocolized, but you still get traffic jams because people are going off in all directions and the system was not designed with particular directions in mind (modulo some slight anticipation of say daily rush hour commutes or physical geography).

Why is the disorganization necessary? It’s because undirected potentiality has very high maintenance costs for higher-order maintenance. You have to keep it ready for many directions all at once. Imagine keeping a kitchen ready to cook any kind of cooking at a moment’s notice, including with the right fresh ingredients. Super costly. So you’re going to let it get disorganized. You’re going to let some things expire and rot. You’re not going to put a lot of thought in how you store stuff. In a word, your organizing principle for the fat is going to be mediocrity.