The phrasal template random acts of ________ is clearly one of my favorites. I seem to have used it 20+ times on Twitter in the last few years. Here are the actual instances:
- random acts of ontology
- Random Acts of Web3ing
- random acts of policy vandalism
- random acts of templing [as in, treating something as a temple]
- random acts of patchy, pointillist, impressionist worldbuilding
- Random acts of philosophy in the “air game” and random acts of tinkering in the “ground game”
- Random Acts of Magical Thinking
- random acts of tariffs
- random acts of sciencing
- random acts of art production
- random acts of revenue-generation
- Random acts of petrichor
- random acts of strategy
- random acts of cash-flow management
- random acts of consulting
- Random Acts of System Integration (RASI)
- Random Acts of Product Development
- Random Acts of Workflow Improvement and Unnecessary Optionality
- random acts of solutionism
- Random Acts of Mildly Profitable or Break-Even Teaching
- random acts of twitter strategy
- Random Acts of Overt Marketing
- random acts of garam-masala-ing
At one point I tweeted a prompt inviting people to fill in the blank, and got a whole bunch of responses, some clever, others not so clever.
iirc, the very first example I encountered, sometime in the 90s I think, was “random acts of marketing.” That stuck with me because it seemed like such an apt description of the marketing efforts of most companies.
Random acts of X are a regime of behavior that you might call “bullshit agency” — some fraction of it works, but you don’t know, and to a certain extent don’t care, which fraction. Hence the famous John Wanamaker quote, ““Half the money I spend on advertising is wasted; the trouble is I don’t know which half.”
Random acts of X happen when you act opportunistically, based on circumstantial possibilities and very little thought, and with indifference to whether or not your actions make any sort of larger strategic sense. The randomness in what the immediate circumstances allow or encourage you to do translates to randomness into what you actually end up doing. Noise in, noise out.
This does not mean that the opposite of “random acts of X” is strategy. You can have “random acts of strategy” too, and in fact most strategy fits that description. A CEO goes off on a leadership retreat with a few buddies, enjoys good food, good wine, and whiteboard sessions, and returns with a nice mind-map and strategy notes… and it’s back to the quagmire of operations within a day. That’s random acts of strategy.
Random acts of X regimes are attractive because they allow you to act in very low energy regimes, with low intelligence. And we default to such regimes as a slightly superior alternative to being frozen in inaction and doing nothing at all. The leap of faith underlying random acts of x-ing is belief in a benevolent universe where doing something, anything, beats doing nothing.
Reviewing my tweets, I notice that I use the phrasal template more often to refer to my own behaviors than to comment on others’ behaviors. The template has no particular stable valence for me. Sometimes random-acts-of-x-ing is good, sometimes it is bad.
But looking at my (over)use of the template, I do wonder, what does it take to move such behavior into a non-random regime, without overwhelming it with the artifacts of deterministic planning, and destroying what little energy there is.
The best guide I’ve found so far is Charles E. Lindblom’s classic 1959 management article, The Science of Muddling Through. It is one of the articles I recommend most often to consulting clients (I found it via John Kay’s excellent book, Obliquity)
Muddling through is the act of adding just enough determinism to a default random-acts-of-x situation to get it to make some sort of roughly right directional progress. In Lindblom’s account, muddling through involves a “method of successive limited comparisons” as opposed to a “rational comprehensive” approach.

Muddling through is both a better term, and a better concept, than its degenerate modern descendants like “agile.” The salient feature of Lindblom’s account is that he doesn’t claim muddling through is a “theory” but rather a manner of doing that “greatly reduces or eliminates reliance on theory.”
Still, whether you call it agile and pretend you have a theory, or call it muddling through and admit that you don’t, the problem remains — how do you prevent this regime of behavior slipping into either useless randomness or getting swamped by the imposition of energy-draining theorizing?
One part of the answer is, as Karl Weick argued, to give up on theory, but not on theorizing. The idea that “what theory is not, theorizing is” has been the linchpin of my consulting work for a decade now, but I’ve never quite clarified the essence of the distinction to myself.
Weick’s idea is similar in spirit to the Eisenhower line that plans are nothing, but planning is everything; or Frederick Brooks’ idea that you should “plan to throw one away” (and Joel Spolsky’s counter-argument that you should not throw one away)
I think the common thread here is that your history of engagement with a problem or question is important, but the specific conceptual scaffoldings you used in generating that history are not. The data matters, the algorithm you used to generate it doesn’t. Be the data, not the algorithm.
This then is the solution to the perils of the “random acts of X” regime — better memory. Turn the memoryless random acts of X into memoryful not-so-random acts of X.
This assumes that memory by itself has something like a gradient to it; a historical logic that can bias the context of random-acts-of-x-ing enough that your actions acquire a drift, a direction of muddling through.
This direction is not a True North. It is not a teleological potential induced by a goal, but an etiological potential induced by a history (or more generally, data). A True Past perhaps. The test of truth being that it creates a coherent future despite the randomness of circumstantial forces. Such an etiological potential is, however, merely necessary, not sufficient. To get past historical determinism, the True Past must only be allowed to frame the random acts of x-ing in the present, not fully specify it. And if your random acts are not capable of blowing up the historical context that contains them, they are not random enough.
I think of it as “fuck around and find out, but never forget.”