The side hustle to my side hustle is helping make the social side of side hustles more fun.
So many people seem to think the collective noun for free agents is “market” as in, Hobbesian gig market with race to the bottom pricing dynamics and ugly ratings manipulation games.
Conversation
Just because free agents don’t free associate through shared employment in paycheck corps doesn’t mean they are lone wolves who either stay out if each other’s way or compete 1:1 in Hobbesian ways. Economic sociability is not limited to markets and corporatized entities.
1
2
25
That said, infrastructure for doing anything other than engage in bidding wars on upwork is very limited. There’s no way to simply form a team to bid for an opportunity beyond trawling your general social network. There’s no way to talk about gigwork beyond vanilla social media.
2
1
20
The “enterprise” is a stack of complex software with way deeper collaboration/workflow capabilities than social media. Free agents cobble together ersatz imitations of this stack from underpowered free tools. We need a gigtech software distinct from both consumer and enterprise.
2
1
28
I wasn’t aware of these problems until recently because in 9 years as a free agent, I did *very* little collaborative work with other free agents. Just farmed out some subcontracts and used a few support people off and on. I used to rationalize this as “I just like working alone”
1
17
Now that I’m actually consciously trying, I think what actually held me back is sheer lack of even the most basic infrastructure. Even just a few months in, the experiment is revealing just how enormous the potential is if we can get the infrastructure right.
1
24
“The fortune in the combinatorial space of free agents”
A group of n free agents has at least 2^n ways of creating value, not just n. And that’s an underestimate, since even the same team can collaborate in many different ways.
Free agents radically unbundle the corporation.
5
9
29
This Tweet was deleted by the Tweet author. Learn more
Yep, you need good ways to navigate it in emergent ways, but the path dependent degeneracy configurations we know are not necessarily the best ones for the present environment. In some work I did, I used reinforcement learning to identify good configs from work history.
Replying to
Are you aware of any formal work on efficiently exploring rugged/hard/complex/highly non-convex value landscapes? I'm especially interested in ways of using abstraction to identify which paths get stuck in dead ends vs. reach optima.
1

