Winning Decisions

4 part decision making process:

Framing
Collecting information
Coming to Conclusions
Learning from Experience

Frames is such a copout - it’s this huge abstraction for every mode of thinking about a situation, every contextual bit of information, ever mental model applied. It’s a truly massive set of concepts. In most of these situations it just looks like “what should we care about when making this decision”, or “how do people think about the decisions they engage in?”

An important question is how efficiently I can run through these decision/execution loops.

I’m going to analyze a fairly rigorous decision process here. Expanding options is also a possibility in software engineering, and is what happens when Ulanov comes over and tells you to use multilayerperceptronparams or to do min/max scaling without params. These are decisions that show up as options when you think about how to design a system well. The frame of ‘just get it to work’ and ‘minimize time spent’ actually end in a worse place.

The time allocation between framing, gathering intelligence, coming to conclusions, and learning from experience is worth thinking about. The other dangerous angle is not realizing that you’re making a decision and coming to conclusions based on an assumption.

Decisions to gather intelligence can also look like learning how a system works before making decisions about how to best interact with it. Refusing to put any time into that can lead to much more wasted time.

I do think that this book is far too theoretical, and ‘educating’ people about how to make decisions in theory and then measuring their change merely proves that people can be convinced of your ideas, not that they think they’re important and worth using after practicing them. We want a process that’s optimized for delivering results, not a process that’s optimized for convincing people that it works. That said, believing can be much more important than actual value (just look at pickup).

I still can’t say exactly what framing entails.

They do entertain hypothetical questions that are consistent with Decisive and the rest of the literature. What is the crux of the issue that I am facing? (This seems worse than useless to me - it doesn’t generate any good ideas, and when it does it oversimplifies its space) How do I believe decisions like this one should be made? (Great commitment to a mode of thinking for the problem) How much time should I spend on each stage - as a first guess? (I worry that this is too meta, and so will have to be deployed sparingly or cut down accordingly based on the situation) Can I draw on feedback from related decisions and experience that I have faced in the past to make this decision better? (This is the default set of assumptions that lead to most decision processes not happening at all) What are my own relevant strengths and weaknesses? (Presumes a team of potential actors)

A reason this process is worthwhile is that spending time trying to solve the wrong problem is extremely wasteful.

They cache out the question ‘what is the crux of the issue?’ as going back to fundamental incentives / the underlying drivers of the decision. Which is an extremely valuable question, and is the basis of goal factoring / option expansion. Weird that they picked this framing of the question.

Asking how to make the decision can lead you to experimental solutions - try a few options and go all in on the best ones, for example. Very different than theorizing and hoping you pick correctly.

Also, the cost to this process isn’t that it’ll take hours to make a metadecision. It’ll take a few minutes. The cost feels large because this has to be done, and doing things is hard in general. I expect events to take up the time they consume plus the hours of waste that go along with it - it makes every commitment feel much more expensive.

The questions on how to decide should be applied to every aspect of my life. How do i make decisions while building systems? How do I decide what systems are worth creating? How do I decide who to be friends with, whom to interact with, what communities to build reputation in, etc?

Making the appropriate metadecision as something that’s essential no matter the time constraints - there’s this general sense that this kind of investment isn’t worth putting in. But it saves so much time that it’s worth it. Say that writing cnns on spark isn’t actually the best decision for me. Doing something else (or many other things) that are productive may actually be the optimal play. Perhaps I should invest 100% of time into becoming a more effective programmer, learning probability theory and linear algebra deeply and taking on small projects that can be started and finished in the space of a week.

Source: Original Google Doc