When setting out on a venture to create something you need to understand lots of things.

There's sort of tried-and-true wisdom about "Budget, Timeline, Features, pick two".

There's a much more nebulous one, that I'll just call "risk". It happens both for writing software and dealing with physical creations in much the same way.

Surely, don't let that stop you, because the best way to understand this stuff is to just go do it, and know that you'll probably screw up a lot. Keep going. Try harder.

Your risk rank-order depends a lot on your other constraints (budget, timeline, features). I think this covers everything from building models as a hobby and inventing new tech. It's one of the most important frameworks I have for decision making.

Let's start with a mythical project that is not meaningfully budget constrained, but has a strong timeline and strict feature set, that feature set is at the bounds of what is possible for you to accomplish, so you're feeling nervous. Maybe it's at the bounds of what is possible for humans to accomplish together, like going to the Moon. Let's take something like that.

If you can return paint that's the wrong color, or over-ordered, then you should buy lots of extra.

Problem Solving, Purchasing and Invention

First of all, if you can't identify your problems, and I mean your core problems, not your symptoms, then you are hosed. This suggests that you need to understand what it is you're doing. If you don't understand it, then your risk is getting dialed up. If nobody on your team understands it, then you have offloaded your risk to powers outside your control.

Here's an almost-contrived example:

  • Symptom: The pre-heater was too cold ever since Monday morning, and nobody could use the forge until 10AM.
  • Naive identification: The system says it should be 458 degrees at 9AM, but it was only 312, must be a bad sensor.
  • Naive solution: "Turn the temperature of the pre-heater up to compensate for a bad sensor"
  • Proper identification: The Sunday before the Monday you noticed was the beginning of Daylight Saving Time, but our vendor didn't account for congress moving it back three weeks.
  • Proper solution:
    • Change the clocks.
    • Reset the temperate value to proper settings, your sensor was fine.
    • Set calendar reminders to change your clocks at both the old and new DST dates for the next couple of years.
    • Get your vendor on the line to fix the core issue.

Your production risk was black-boxed into a solution a vendor sold you. That's maybe for the best. If you had relied on something an engineer or scientist or, worse off, you invented, then your black box is that person. Your risk is that you can still get them on the line, you can get them out of bed, they remember and understand the core problem. They have the tools to fix it...

Everything that you use, invent, make, or buy in a complex system has a risk value. Every material, process, and combination. Good technology gets de-risked over time. We pretty much know the operating parameters of, say, a given bronze alloy. That's good technology de-risking. Similarly, we know the operating parameters of magnetic tape, or many kinds of batteries. Apple slowly de-risks using CNC machines to produce their enclosures over several smaller products before they use the process in the iPhone.

If you invent a thing, or process from scratch, you have to handle all the de-risking.

This is a blog post and not a book, so consider this abridged, but here's a rough approximation of my decision tree. I've also just thrown all vendors into "trusted" and "untrusted", without any guidance on how to establish trust. Good luck! I might write more on that later. I'll also point out that this is only most rigorous when the cost of failure is high.. If I'm building shelves or party favors I just go pickup whatever happens to be at home depot, or sitting around my workbench.

  • Can you buy this from a trusted vendor?
    • What happens when something goes wrong?
      • Can you look under the hood and fix it?
      • Can they?
      • Will they?
      • Who pays?
    • Will it arrive on time?
  • Can you buy it from an untrusted vendor?
    • Same questions, but mentally quadruple your risk factor.
  • Can you invent it?
    • How long will it take?
    • How much will it cost in materials?
    • How long and what tests will you do before you decide it is fit for use?
    • Quadruple all of your estimates.
    • Have you done it before?
      • How many times?
      • For how long?
      • What went wrong?
      • How long since something failed?
    • Is it close to something you've done before?

This has a lot of implications for other decisions though.

Don't pre-order shiny complicated items (electronics) from internet campaigns with far-future (3 month+) shipping estimates. If they ship they'll be late. You probably don't want the first version anyway.

Risk is like interest, it compounds, the more complex a system is, the higher chance of failure. Keep pre-ordering socks and comic books and art. stop kickstarting supercomputer ovens. Let the corporations manage that risk, not your wallet, that's what a C-Corp is supposedly for.

There are exceptions that prove this rule. Trust Bunnie.

As an aside, and maybe a later post: Sales is the exact opposite process. If you're selling something to somebody, from toothpicks, to multi-year service contracts, your job is to convince them that the thing you have is the least risky way to handle their problems. Make a list of all the risks they have about you, your product, your process, your team, and be ready to have an answer for why those are assets and not liabilities.

Comments and Messages

I won't ever give out your email address. I don't publish comments but if you'd like to write to me then you could use this form.

Issac Kelly