Introducing MicroCommit
What is the smallest practical commitment an organisation can ask of its members?
This is an article introducing MicroCommit, a simple website where you can register and get weekly requests and tasks related to preventing extinction risks from superintelligence sent to your inbox. The only expectation is for you to acknowledge the requests and tasks by explicitly accepting or rejecting them, instead of just ignoring them.
I think coordination is very important. Whenever we collectively fail at a large-scale endeavour, I usually diagnose some form of coordination problem.
Whenever I meet people who agree with me on the importance of coordination, they sadly tend to focus on topics I find quite irrelevant in the context of solving coordination problems.
Either they are too theoretical, like mechanism design and game theory.
Or too spiritual, like Buddhism and Spiritual Enlightenment.
Or too intractable, like “improving the culture”.
Prosaic Coordination
The coordination problems I have faced in the past are instead very prosaic.
Here are some examples…
People avoid regularly doing boring things. In many groups I have been part of, regular requirements like writing activity reports or attending meetings on time, have been systematically neglected. The few groups that have made these things happen tend to have put a lot of energy into it.
People avoid conflict. People constantly avoid people who contradict them, and contradicting people. Sometimes, this leads to groupthink, where everyone thinks the same. Usually, this simply leads to dysfunction: people still disagree, but just stop engaging with conversations.
People are not reliable. If someone tells you “I’m going to do [a thing]”, they usually won’t. If someone asks you “Can you please do [a thing for me]?”, they usually will not ping you about it if they still expect you to do it, nor will they tell you if it’s not relevant anymore.
—
Nerds have a natural aversion to coordination. This one is specific to nerds, but I am a nerd, work with a lot of nerds, and expect a lot of my readers to be nerds.
But yeah, nerds have a natural aversion to coordination. Coordination requires going against a lot of nerd instincts. Doing things that one does not fully believe in; doing things that one would not have done by oneself; or deferring to consensus instead of one’s own beliefs.
As a result, nerds avoid some common failure modes: nerds are (usually) not political extremists who scream at people online. But they also develop new failure modes! Like making it impossible to agree on courses of action not fully specified ahead of time.
MicroCommit
MicroCommit aims to solve one specific coordination problem.
How can an organisation, at scale, reliably ask its members for help, and know who will reliably consider helping?
This is a very common problem. Organisations constantly want help from their members: boosting some piece of social media content, stating what they think of a policy proposal, participating in a poll, buying a book, etc.
Right now, there is no reliable way for organisations to ask for this help at scale…
Social media is not the place. Organisations have no guarantee that their posts will reach all of their members.
Newsletters are not the place. It’s very hard to distinguish “I have seen the request and plan to not do it” from “I have not even seen the request or did not consider that it applied to me”
—
This is the problem MicroCommit solves.
The solution is to formalise the notion of “a very small commitment”: every week, taking 5-10 minutes to review requests from an organisation, either accepting or rejecting them, but never ignoring them.
This way, organisations can just see the number of people who reliably (let’s say, over the last 2 weeks!) consider their requests.
Because of my experience with preventing extinctions risks from AI, this is where I will be starting.
ControlAI has a lower engagement way of interacting, namely through subscribing to its newsletter, and I hope MicroCommit can get at least 1-10% of the newsletter’s subscriber count.
Conversely, Torchbearer Community requires a much higher commitment than MicroCommit, and I hope MicroCommit can get at least 50-500x of TBC’s member count.
If MicroCommit works well, I hope that over the next few months, other organisations will be interested in using MicroCommit as well. And if enough do (like 5-10 serious organisations), I’ll likely open MicroCommit for them too!1
—
The current alternative to MicroCommit is for organisations to do a full campaign for each specific request.
For example, if an organisation wanted to ask its members to buy a book, it would need to talk about that specific book for many newsletter emails, change their banners on social media, ping everyone on their community servers (like Discord), et cetera.
Scheduling a full campaign every time an organisation wants to secure the attention of its members is way too expensive.
Organisations would tremendously benefit from the ability to cheaply request 5-10 actions from their committed members every week.
My Hope
My hope is that this can make a dent into the race for cheap attention that social media currently favours, and reward organisations for focusing on people who are more active.
Ideally, I hope to see possibilities for from organisations to move to more substantial engagement requests than “Please QT this!” or “Contribute to this Patreon!”
I want organisations to more often schedule events, debates, giving feedback on proposals for actions, and more generally offering opportunities for short-term volunteering.
Right now, each of these requires a lot of investments, and by lowering this cost, I hope it becomes more common.
Conclusion
I hope you like the idea, and join MicroCommit.
If you care about the future of AGI, coordination and the world, and have 10 minutes to spare a week, please sign up today!
(You may also leave feedback in the comments, or on the form.)
Cheers!
An old version already supported other organisations managing their requests. So in theory, once we’re finished with fixing bugs during the beta, it should be easy to add the feature back :)