Do Things That Work

by Aaron on January 10, 2012

In my junior year English class we’re discussing Nietzsche’s Ubermensch. One student asked the teacher “Wait, so if you’re an ubermensch and you come across a traffic light that you don’t like, do you just run through it with a monster truck and squash all the cars in your way?”


The class laughed. The teacher said yes.


Later on, I asked what would happen if you were an Ubermensch, but you didn’t have a monster truck.


“Well, ” my teacher said, “then you die.”


If you’re going to try doing something, you have to do it in a way that works. A lot of people try to copy something that looks successful, without learning why it works. Even more don’t realize that they don’t know how it works. This causes lots of problems.


It’s the person who decides to study, getting distracted by facebook, then spending lots of time and not getting results.


It’s all the people who go to the gym, eat crap, and don’t get any stronger.


It’s Iceland deciding to do finance, starting a bubble, then crashing the economy.


It’s really easy to look at people’s surface behavior, then try to copy it, but leave out all the parts that actually work.


Drowning looks a lot like swimming. You flail your arms around and kick a lot, but you don’t push against the water enough to keep your head above the surface.


Where does the push come from? What makes things work?




When I succeed at something, it’s for a reason. I can trace a chain of cause and event between what I did, and what I got. The things I’m best at are the things I understand the most.


I can build a robot because I can see what every damn part is doing. The mechanics don’t hide anything — gears have teeth that force other gears in a circle. Mechanisms rotate around hinges. Things hold up because metal would need to break if they didn’t.


Electronics is trickier, but if you have time, a multimeter, and debugging LEDs on your mechanisms, it’s not that bad. If you can swap out everything you didn’t build, it’s pretty easy to isolate the problematic part and remove it.


If only everything were like that. Interacting with non-physical facts is harder.


If you don’t want to be blind, you need metrics. That way, you can see what’s happening. The metrics don’t need to be super fancy, but they do have to be related to the thing you want to track.


Worst case scenario, you can science it out. Try something, see how it affects your metric. Try variations of the things that work out better. Local maxima can happen, but at least you’re doing better than random.


But if you really want something to work, get the causality straight. Figure out why people are responding the way they are, and amplify that. Pay attention to comments. Talk to your readers.


One thing that’s interesting about working with Sebastian is how unapologetic his stances are.


When we were putting together the book, the team kicked titles and subheadings back and forth. Sebastian told us to be less beggy and more direct before, so we were coming up with different titles. Suggestions like “Warpath” or “Letters from an Imperialist” came up. You know, fun aggressive stuff. We were having a grand old time.


Then Sebastian replied.


“No no no! If we talk and market like that, we’ll get torn up. You all want STRENGTH STRENGTH STRENGTH in the title, but trust me here, that won’t actually work.”


I was just trying to pattern match to Sebastian’s previous assertions, rather than following the chains of causality supporting them.


In all the cases he wanted us to be decisive, we already were in a good position. We had something valuable, and we were trying to reach people would would like it.


Isn’t it annoying when people just pattern match, and don’t do things that work?


They look like they know what they’re talking about, because they’re using the same words that people who do are using. They don’t know they don’t know how it works, because they’re saying things that make syntactical sense.


But that’s no good.


{ 1 comment }

How to Move Time Within OODA

by Aaron on January 6, 2012

I like thinking about timing, and I’ve recently done a little bit of thinking about marketing, so here’s my all-in post on timing. If you want me to continue writing on the subject, share this with your friends, because I’ll start focusing my attention elsewhere unless this goes awesome.


And without further ado…


OODA is cool. Timing is cool.


You observe what’s going on, orient yourself to figure out what it means (and reevaluate your observations), decide what to do, act, and then look at the results.


There are lots of implications, and ways that you can play with it. I’ll talk about the ones I learned about.


How frequently you can observe things changes how frequently you act.


If you go through an OODA loop every time you act, you can’t act without observing, unless you want to shoot blind. More on that later.


One of the biggest differences between now and about this time last month is the pace of the project. And I think that Observation speed is a major cause of that.


When we were putting together the book, I could get feedback on the organizational structure by just rereading sections in the new order. This would take like, 10 minutes. I could get faster feedback by skimming, and cut that down to minutes. By the time the book was over, I could recall posts on the order of seconds.


If I wanted someone else’s opinion, I could talk with them about it. If I wanted them to directly observe the book, this would normally take a few hours. If I skipped to sharing my Orientation, I could get feedback from other people in like half an hour if they were online.


Now, we’re trying to build relationships with allies over email, and write posts to get people to come on the site and build awareness.


This takes a while. Like, as a rule of thumb I expect an email with someone to take like, a day, unless I respond to them within an hour.


Still though, things seem to be maxed out at one meaningful Observation a day. It takes a while to find out if a post works, or if something is scheduled, and it takes a while for things to come up. So the project slowed down.


You can precompute steps.


I previously talked about speeding up your actions by cutting down on Orientation and Decision making.


Sebastian is really good at this. One of the most interesting things about working with him is that he really quickly has an interpretation of what’s going on, even if he doesn’t have that much information. For instance, everyone on the project was mailing back and forth about title names and different marketing angles and considerations, and we kind of didn’t know what would work. We had reasons for thinking things, and we would need to test them out. Then all the sudden Sebastian would come in and say something that made sense, and we would all change our minds.


Basically, he’s able to decrease the amount of time needed to Observe and Orient.


But he’s not pulling it out of his ass. Most of the time, he’s right about what’s happening and what we should do, or at least more right. Posts with his influence do better. His marketing efforts get farther.


But you still need information in order to accurately know what’s going on. You can’t covary without getting bits in. Where do his bits come from?


Principles. Where do those come from?


Experience. He just has more of it than we do. He’s done lots of projects, and reads ridiculous amounts of history. He already has a lot of information and Observations about things that happen, and has already Oriented himself with regards to those.


He now gets to reuse Orientations. Assuming that things work similarly to the way he did with his other observational data, he needs less time to Observe and Orient.


Living on Principle basically just lets you decide everything faster.


You can trade time between steps.


There are 24 hours in a day, and you can never, ever, get any more.


But you can change how many OODA loops you get in a day.


For instance, you can spend less time Orienting and Deciding, fire more randomly, but be able to cram in more observations. This is basically the “fire bullets, then cannonballs” strategy.


If your Orientation or Decisions aren’t very good, then this is probably what you should be doing. If something works, you just need to figure out why, then repeat it.


If you’re inexperienced and thus your Orientation and Decisions are a bit funky, then you really should be doing this. It gets you the Observations you need in order to build up a framework of what’s going on, and doesn’t overfit to your previous training data.


If you have lots of experience, you can just live on Principle, and see how well they work. Combining strong principles with lots of shots, and you can get success and more data at the same time.


You can substitute steps.


Sometimes though, the Observations limit your speed.


For instance, a decision might need input from someone else. What can you do then?


What I did a lot was trade off Observing their actual response and spending more time Orienting myself to possible answers. I would try and predict what they would say, then tell them that I was guessing that in the email.


If my confidence of my prediction and the utility of the action were both high, then I could just act on that default assumption. If it was correct, then yay, if not, then oops.


Another possibility is just to consider both answers, and do what you would do in either case.


Observations can substitute Orientations or Decisions by just letting you try lots of things, then see what what works.


Orientations can substitute Observations by using previous data.


Orientations can substitute Decisions by filtering out information. (This one seems riskier than others)


Decisions can substitute Observations via precommitments.


Playing around with this is left as an exercise to the reader.



An OODA Perspective on The One Week Book

December 22, 2011

Originally, the plan was to find organize core posts and auxiliary posts into chunks, those chunks into chapters, those chapters into sections, and those sections into a book. This was fine and dandy, but it basically failed to account for that it assumed that I would have any idea what would be turned into the book. […]

Read the full article →

Keep your Balance

December 13, 2011

I was hiking in California with some friends, and Jasen Murray said something to the effect of When you’re climbing, you must only keep your balance. It doesn’t matter if you’re at the top of a cliff, or on a branch, or on the ground — the height can’t make you fall over, only losing […]

Read the full article →

For the Web, By the Web

December 7, 2011

The Publishing Industry is ripe for disruption. They’re too slow, they don’t treat people well, and they’re based on monopolized marketing channels and obsolete technologies. The internet lets people collaborate and distribute at high speeds for low costs. Bloggers generate lots of great content, and social media allows for actual word of mouth marketing. We’re […]

Read the full article →