AI doesn’t solve Work Theater
TL;DR
Big companies fail by building the wrong thing, not by building too slowly. The author predicts AI coding will bring small or negative gains to companies with more than four management levels.
The article
- Work theater is the pattern at large companies where projects that look meaningful in Jira crowd out work that delivers something real to customers.
- "Think less, act more" works only once you have product-market fit and know what to build. The rest of the time you should slow down, ask more questions and hear from more people.
- AI speeds up the 0-to-1 build. It does nothing for deciding where to take a product next. It also makes it easier to fake progress with a flood of tickets and changes.
- The author can't name a failed big company, Yahoo included, that was building the right thing 20% too slowly. Failure almost always came from picking the wrong direction. Often employees and customers said so at the time.
- The author cites "Recursive Middle Manager Hell" and his own dozen jobs to name management levels as one of the biggest sources of waste. He predicts companies with more than four management levels will get small or negative returns from AI. The negative case comes from AI letting optics-driven engineers pile tech debt on everyone else.
- Companies with three or fewer levels should speed up a lot. The author welcomes that competition, which forces incumbents like Meta to improve or be replaced.
Why it matters
It's a two-minute opinion piece with no data beyond the author's own experience. Read it for the specific prediction about management levels, which is easy to check against your own company.
The discussion
A small, mostly agreeable thread. Commenters accepted the premise that AI feeds work theater, and several thought it will make things worse. The pushback went to the article's second half, which blames middle managers.
AI will make work theater worse
The most-replied comment predicted a golden era of work theater, and one reply said it has already started. Another joked about accepting a token-burning award. One commenter hoped faster building would reach a product's dead end sooner and force companies to think, but expected more doubling down in practice. A reply argued that theater was never about profit, only make-believe, pressure and ego.
If anything I’d guess that AI would usher in a Golden Era of Work Theater the likes of which we have never seen
In reality, I have a feeling that it would lead to even more encroachement, more doubling down and less useful work.
Blaming middle managers misses the point
Two commenters rejected the management-levels argument. One said good managers keep a team on work that matters, and bad ones build empires. Another liked the opening but found the turn to blaming middle managers for everything jarring.
If anything, good management incentivize the right things and focus a team on work that matters. Bad managers only try to empire build and incentivize the wrong things in their engineers.
I think that the start of the article was interesting, then it goes down the path of middle managers being the root of all evil, just seems a bit like whiplash.
Cheap building removes the reason to say no
One commenter agreed that building the wrong thing causes most failures, and said AI has made it worse. Cost used to be the reason to turn down a feature, and now teams say yes to everything. Another added that AI overengineers by default, so the bloat piles up.
Historically, that answer was ROI - why spend days/weeks/months on something that might not be valuable. Now I see people just saying "Yes, and" to all feature requests.
If anything, AI tends to vastly overengineer things (in a bad way) which makes bloat inevitable.
Consensus
AI speed doesn't fix building the wrong thing, and it probably adds to the busywork.
Dissent
Management levels aren't the cause. Good managers keep teams on work that matters.