The Trouble With Babbles
post by AllAmericanBreakfast
I haven't yet participated in any of our babble challenges, even though I think they're a good idea. My reason is that they are ambiguous and prone to Goodhart's law. If the challenge is to list 50 ways of getting a golf ball to the moon, no matter how silly, I could do this:
- Carry it while jumping to the moon after getting really buff
- Carry it while jumping to the moon with rocket boots
- Carry it while jumping to the moon from a high platform
- Carry it while jumping to the moon from a high ladder
- Carry it while jumping to the moon from the top of a human pyramid
You can see that I'm optimizing here for getting to 50 by proposing minor variations on the same core idea. The intention of babble is to get you out of your mental ruts and past your blockages. It's not clear that this exercise on its own helps you do that very much. Yet if I said to myself "no more 'jumping' entries," I'd be pruning.
Can we solve this problem, while keeping the spirit of the exercise (which itself is a commendable example of babble)?
Let me babble a few ideas:
- Every 20 babbles, go back and categorize your ideas. Make the cutoff "At least 50 babbles, and 10 categories."
- Try problems in a conjecture + counter-example format. Topics that might work well here are social, philosophical, or aesthetic.
- Choose problems that are not technical problems (getting an object to the moon), but personal/social problems ("babble 50 ways to incentivize better teaching").
- Allow respondents to choose their own problem.
- Babble baseball. One team's "at bat" and they have to babble ideas. The other team's "in the field" and they have to categorize (or sub-categorize) the ideas. Then they have to propose their own ideas within each category. The "at bat" team has a 1-minute head start to list some ideas, which they publish all at once. They then continue publishing new ideas while the "in the field" team goes to work. The trick is that the "at bat" team has to publish ideas sequentially - i.e. "batter 1" has to publish an idea before "batter 2", "batter 2" before "batter 3" and so on. Only when the last "batter" has published an idea does "batter 1" get to go again. By contrast, the "in the field" team gets to work all at once.
Please add your own babbles in the comments!
Comments sorted by top scores.
comment by Elizabeth (pktechgirl) ·
2020-10-22T23:38:32.658Z · LW(p) · GW(p)
6. Draw ideas instead of writing them down in words, so there's more time to think between ideas, a higher barrier to committing them to paper, and you'll get bored with very similar ones faster.
comment by AllAmericanBreakfast ·
2020-10-23T00:04:45.080Z · LW(p) · GW(p)
I like this a lot! It dovetails with all my writing and introspection on visualization and learning, which has been the most important insight for my learning process this year. Concept handle: “visual babble” or “visual brainstorming.”
comment by gjm ·
2020-10-22T23:34:55.719Z · LW(p) · GW(p)
I have the same concern. My own solution, to whatever extent it is one, is to try to make my answers reasonably diverse and creative -- which, yes, means doing some pruning -- but also to allow myself to write a boring answer or two if it seems like I am not going to make it to 50 without doing so.
I might rationalize this by saying that the actual skill that it's useful to train is not "babbling" or "pruning", both of which are implementation details, but "producing lots of potentially useful ideas", and that even taking some trouble to avoid boring answers I still find myself erring on the side of not-really-useful ideas, which suggests that whatever the underlying mental process I am at least doing something babble-ish overall.
(But that would be a rationalization. I think the actual reason is that writing boring things feels unsatisfying.)
comment by AllAmericanBreakfast ·
2020-10-23T05:03:22.551Z · LW(p) · GW(p)
The original Babble and Prune article at first acted on me almost like a magic creativity drug. That wore off, of course. But at first it seemed like a cure-all for blocks. However, I know from past experience that if I just let myself write aimlessly, I can produce an enormous volume of pointless crap that contains zero good ideas. This general sort of thing has been tried by many, many, many writers in the past: stream-of-consciousness, getting drunk, arbitrary rule-systems for making choices, etc. The "50 ideas for X" is another one.
And they're probably not bad practices, but they're also not guaranteed to produce great literature or great anything else.
I'm more interested in practices that sort of merge or dissolve the dichotomy between "babble" and "prune."
One that worked for me when I had time to do creative writing was to write a single word or phrase, then ask, over and over again as I proceeded, "what question might my reader ask about this?" And then to answer it as directly as possible.
This was nice, because it was open-ended enough to let the ideas flow, but specific enough to give them form. Here's a brief example off the top of my head:
He ["who is he?"]
was a shepherd ["where did he live?]
in Iceland ["when?"]
for about three weeks back in college. ["What brought him to Iceland?"]
His uncle Gunnar invited him to visit the family plot ["why?"]
after his wife died suddenly ["of what?"]
of what the postmortem determined was an aneurysm, but Gunnar claimed was because she accidentally said a kind word to him that morning. ["Hey, what's the main character's name, anyway?"]
Eric ["Gunnar sounds like a jerk, why would Eric want to spend time with him?"]
, caught in the terminal stages of a relationship on the verge of metastasizing to his legs and his heart, used the opportunity to flee the country back to the land of snow and wilted, overpriced vegetables.
comment by Slider ·
2020-10-23T14:04:37.202Z · LW(p) · GW(p)
In my mind I solve this issue by understanding that it is 50 distinct things. While there is wiggleroom and it can be subjective, asking "Is this distinct?" makes you recognise whether you are lying to answer "yes".
One could have a exercise where the challenge would be to describe an interesting approach in very high detail. But in this style you look for alternate and separate ones. For example "Carry it with me while jumping off a high support" woudl encompass 3,4 and 5. But for example "ask NASA for a trip to the moon" would not be a duplicate of 5 despite one using a group of people as we are using it as a social strcture and not as physical support structure. That is thinking how the entry answers the question limits and frees up possible answers.