Training Regime Day 7: Goal Factoring
post by Mark Xu (mark-xu) · 2020-02-21T17:55:29.848Z · LW · GW · 6 commentsContents
Introduction Cleaning your closet Steps Example: Learning Chinese Exercise None 6 comments
Introduction
Humans have these things that we call goals. They also take these things called actions. Ideally, a human takes actions that move the world closer to one of the human's goals. You want your actions to be connected to your goals in this sense.
Usually, when an action is first introduced into your action repertoire, it has a very clear connection to your goal. However, once an action is introduced, it often tends to stick around. Sometimes, the action-goal connection doesn't change, and so continuing to take that action is a good idea. Other times, the action-goal connection shifts, but you still continue to take that action.
The purpose of goal factoring is two fold. The first purpose is to evaluate the strength of the action goal connection for any given action. The second purpose is to find actions that have a stronger connection to your goals.
Cleaning your closet
Closets can have many things inside them. The closet I will discuss today contains all of your goals. However, you can't just put stuff in a closet willy nilly, it must be organized in boxes. The boxes that you put your goals in are your actions.
We proceed as if we are cleaning a regular closet. First, we pick a box. Then we open the box. Sometimes, we realize that the box is way too big for the thing inside it, so we find a smaller box. Sometimes, we don't even want the stuff, so we throw it away. Other times, we find that the stuff fits pretty well inside the box, so we leave it as it is. Goal factoring is the process of opening the action-box to find the goals inside.
In terms of action-goal connection, the first case corresponds to a case where the action is an ineffective way to achieve the goal. For example, suppose my goal is to stay up to date with current events and my action is browsing Reddit for hours a day. It seems evident that there are ways that will keep me both better informed about current events and take less time than browsing Reddit.
The second case corresponds to goals shifting as time passes. For example, I used to value having up-to-date information about Magic: the Gathering, resulting in me checking various MtG websites quite frequently. However, recently I discovered that I no longer really cared what was happening in the MtG world, yet I continued to check the websites.
The third case is when the action-goal connection remains strong. For example, I have a goal of not having my teeth fall out and I take the action of brushing my teeth. It seems to me that brushing my teeth is one of the best ways to not have your teeth fall out, so I continue to brush my teeth.
Steps
Here’s a diagram I stole from the CFAR handbook about a good visual way to follow these steps
- Pick an action.
- Write down all the goals that the action moves you towards
- Many people have goals that aren't quite socially desirable. It's important to write these too.
- For example, one of my goals for "going to the gym" is to become more attractive.
- Completeness check: check if you've written down all the goals is to pretend that you already have everything you've written down in abundance. If you've written down all the goals, then you should feel no desire to perform the action any more.
- Many people have goals that aren't quite socially desirable. It's important to write these too.
- Write down all the goals that the action moves you away from
- Again, it's important to write down everything, even stuff like "makes me look like a weirdo" and "don't like talking to people".
- Removal check: check if you've written everything down by pretending that the action doesn't move you away from any of those goals. If you've written everything down, then you should feel like taking the action is correct.
- Remember that things like "takes time" are goals that almost all actions move you away from and that you should write those things down.
- Decide if the goals you've written are crucial to that action.
- This can be done by seeing how much your desire to take the action changes if you varied the magnitude of how much the action affected that goal.
- In math language, a goal is crucial if the derivative of action desire w.r.t. action goal connection strength has large magnitude.
- Example: "takes time" is probably not crucial to "brushing your teeth" because I would still brush my teeth if it took double the time.
- Example: For me, "looking better" was not crucial to "going to the gym" because I would still go to the gym even if it made me look slightly worse.
- Example : For me, "acquiring new experiences" was not crucial to "getting my ears pierced" because I wouldn't extremely want to get my ears pierced if the experience was twice as unique.
- This can be done by seeing how much your desire to take the action changes if you varied the magnitude of how much the action affected that goal.
- Brainstorm other actions that achieve the goals you've written while avoiding the costs.
- This is an exercise in [babbling] (https://www.lesswrong.com/s/pC6DYFLPMTCbEwH8W/p/i42Dfoh4HtsCAfXxL [? · GW])
- If the box you open has many goals inside, you can replace the single action with a lot of smaller action-boxes
- Imagine taking those actions instead of the action you're currently goal factoring. Are there any other goals that those actions move you away from? Are there any goals that those actions don't move you towards?
- Decide which set of actions to take, iterating (3) if necessary.
Example: Learning Chinese
My parents speak both Chinese and English. My extended family pretty much only speaks Chinese. My current level of Chinese is passable, but sorely lacking. As of 2 weeks ago, I had spent the past month or so studying Chinese for 10 minutes every day. Here's what I goal factored:
Goals learning Chinese moved me towards:
- Parental approval
- Communicating with extended family
- Want to be able to think in a different language
- Social status boost from being bilingual
- Career prospects increased from being bilingual
Goals learning Chinese moved me away from:
- Takes time
- Takes attention
- I don't like learning languages/memorizing things
- I don't like doing things I'm not good at
After considering which goals were crucial, the list ended up looking like:
Goals learning Chinese moved me towards:
- Communicating with extended family
- Want to be able to think in a different language
Goals learning Chinese moved me away from:
- Takes time
- Takes attention
Brainstorming ideas, I noted that "thinking in a different language" could be achieved by learning a non-Chinese language. In fact, it seemed like learning a language like ASL that uses a different channel of communication might be much better at achieving this goal than learning Chinese.
However, since my extended family pretty much only speaks Chinese, it seemed like "learning Chinese" was the best action to accomplish that particular goal. At this point I just had to weigh the time/attention it costs to learn Chinese versus how much I wanted to be able to communicate with my extended family.
This is what I ended up with:
This process is a lot like just writing a pro/cons list. Although plain pro/con lists are more useful than people give them credit for, I think that the crucial addition is trying to figure out different actions to take to get what you want.
Remember, the point of goal factoring is not to pick and action and convince yourself that it's a good/bad action. Keep your bottom line [LW · GW] empty.
Exercise
Pick an action and goal factor it. Places to look for actions that might be good to goal factor are actions that you started taking a long time ago, actions that take a lot of time, and actions that you don't often think about because they're very small. The latter case is relevant because lots of small actions can pile up and end up sucking away a lot of your time/attention.
Remember to build form [LW · GW]
One way to think about applied rationality is trying to achieve your goals while staying agnostic about the strategy. Goal factoring is a tool that helps you do that.
6 comments
Comments sorted by top scores.
comment by romeostevensit · 2020-02-21T20:07:47.235Z · LW(p) · GW(p)
> This process is a lot like just writing a pro/cons list. Although plain pro/con lists are more useful than people give them credit for, I think that the crucial addition is trying to figure out different actions to take to get what you want.
Good point!
I think of this as pros and cons are reusable between goals, and it's worth learning the general sorts of structure that pro and cons (and their generation) have. Doing this, your sense of the 'pro con space' and how it connects to your longer term goals will improve. You'll find yourself making more modular choices such that overall there is less wasted motion when it turns out you need to modify your sense of the goal or method. In the pedagogy literature, a lot of this falls under the heading of 'multifinal goals and means.' This also suggests a complementary practice of method factoring.
comment by Pattern · 2020-02-22T21:46:47.747Z · LW(p) · GW(p)
General:
I've seen other discussions of this material, but the 'make sure not to do this' parts made it felt more complete:
Remember, the point of goal factoring is not to pick and action and convince yourself that it's a good/bad action. Keep your bottom line [LW · GW] empty.
Style:
Completeness check: [the way you] check if you've written down all the goals is to pretend that you already have everything you've written down in abundance. If you've written down all the goals, then you should feel no desire to perform the action any more.Replies from: mark-xu
↑ comment by Mark Xu (mark-xu) · 2020-02-23T04:27:51.961Z · LW(p) · GW(p)
I agree that that comment didn't really add that much. I was just trying to caution against the view that goal factoring was a technique for convincing yourself to take/not take certain actions. I'm not sure whether I should have spent more time discussing that though, because I'm not sure how common such a failure mode is.
Thanks for the style pointer!
Replies from: Patterncomment by lambiguo · 2020-03-21T23:54:45.427Z · LW(p) · GW(p)
I feel like this idea has the added benefit of getting to know oneself better. I think that some of what I tell myself of my preferences is not really the truth. The idea of analyzing if a quality is crucial by mentally changing its magnitude and thinking about the effect is especially interesting. The completeness and removal checks are also really cool ideas.
As an example, I was looking at 5 man soccer. I really enjoy playing 5 man soccer. When evaluating this, my diagram looked something like:
Inputs: Time, energy, money, organizational effort, interference with other fitness endeavors, effort
Outputs: Exercise, fun, socialization
And looking at those goals this seems like a terrible exchange. I can get much better return on investment via other means if my goals were those. I'm not socializing much at the game. I can have more fun doing other stuff. I can keep in shape more efficiently by other means. But, I still felt uneasy about the decision to quit. That's when I realized that I was failing at the completeness check. What I really get out of this activity is a mixture of: possibility of being competitive in a low stakes environment, primal animal team cooperation effort and vanity in seeing myself improve.
Thus, after all this, I don't think that I will change my behavior, but I understand it much better than I did before.
comment by Conor (conor) · 2020-09-14T00:36:07.726Z · LW(p) · GW(p)
Jacob Fisker has a method called the reverse fishbone diagram.
You draw a horizontal line and that is the action.
Above the line you draw a diagonal forward slanted line for each positive first order effect of taking that action and below you do the same for negative effects.
On those initial branches, you branch off second order effects, up or down pointed depending on their valence until you have a sketch roughly resembling a fish skeleton with as many orders of effects as you can come up with.
You then count the upward and downward lines and compare how the effects serve other goals you have to determine if this is a good action to continue. Of course, how you weight each effect matters, so you can try to take that into account maybe by bolding important effects.
I prefer this method because it is closer to comprehensive by including more consequences in a slightly more elegant manner than the mind map approach.
I'm assuming there is a goal evaluation lesson coming up, so I won't comment on confusing means/actions with ends/goals.