There’s a story to tell about the last century’s history of work: we started measuring input, we got smarter, and we ended up measuring output as a result.
Taylor with his stopwatch. The factory floor where the unit of management was the minute. And then knowledge work arrived, the minute in itself stopped meaning anything, so we unlearned the clock once again. Results-only work environments. Deliverables. I don’t care when you work, I care what you’ve done!
And this was progress. With good reasons. We had stopped confusing motion with value.
But now, given the current times, I want to reevaluate a lot of these assumptions. Maybe we measured output because it was easier to compare to progress? Maybe we started looking for output as a measurement because it was so expensive? And maybe, input, full circle, is again the thing to measure first.
Output as a costly signal
Output as a signal for what is harder to measure in effort and time. Just as Michael Spence fairly simply idea: signals are only informative if they’re hard to fake. Giving the example of why a degree tells an employer more about what went into getting the degree than per se what the curriculum of that degree entails. The cost (effort, time and loans as well) is information and forms a hard to ignore signal.
Output still works the same way. A finished market analysis, a working prototype, a legal memo, a signed contract, etc… are in the same way credible precisely because producing them required something scarce. Time. Skill. Attention. Sustained effort you couldn’t borrow.
So when a manager says “just show me the result”, they quickly jump over the hard part: what went into this? So the result is the proof of effort.
Now that way of thinking could become flawed.
When the signal gets cheap
Put ten agents on ten problems and come back in two hours. You’ll have ten artefacts. A competitor teardown, three landing page variants, a migration script, a cleaned dataset, meeting notes that are actually accurate.
Not brilliant, yet. But in recent months sometimes already close to Pareto efficient , this is, good enough that improving one dimension costs you another. Or at least, these outputs give you a head start that is, when prompted and contextualized right, well on its way.
The argument I’m trying to make is that the point isn’t even that the work is always good. The point is that its cost has collapsed, and with it, its value as a signal.
On one tangent, we could think about what Akerlof described in 1970 with used cars: When buyers can’t distinguish a good car from a bad one, they price everything at the average. Good sellers might withdraw. The market degrades. Now replace cars with documents, reports, ads, creator content, phone calls, … . When any inbox can receive twenty plausible strategy memos in an afternoon, the memo stops carrying information. Not because it’s wrong but because it’s cheap. Recipients start discounting everything. And the people doing genuinely careful work find they have less ways to prove it, because the artefact rather looks identical. Let’s be very aware of this. We call it AI slop now. Only the ‘slop’ sometimes becomes just as good. Can we figure out how to distinguish good sellers? A clear challenge ahead.
On another tangent, you can also see how the measurement of output is not a valid signal anymore. You want to figure out how your employees are managing input. How are they using their time to create valuable input. That is the output to measure. Good input of an employee is the output we want to measure. Still makes sense?
A different kind of input
The input that returns is not hours-at-desk. It’s something closer to capital allocation combined with efficient/productive time management.
When you write a prompt with real context such as the actual financial model attached, three competitors named, the audience specified as a banker rather than an investor, five validation questions decided in advance , you’re building ahead. That prompt might become reusable, shareable, improvable. It sure will be saving you money.
The two hours of compute you spent is a variable cost you allocated. Badly or well. Every knowledge worker is quietly becoming a very small capital allocator in that sense.
The questions that actually discriminate now are embarrassingly concrete. How long did this take you, given that the floor is two hours? What context did you assemble before you started? What did you deliberately not generate? How did you know it was right?
The part nobody wants to say out loud
We don’t just use effort as a measurement. Effort is how you earn the right to the thing you made. We are used to mix our labour with the world so it can become ours for a little bit, for a little while. It’s why “I worked hard for this” is as much a moral argument than a factual one.
But when someone produces in two hours what used to take a week, reactions can go to ‘that doesn’t count’, as if it is unearned.
That instinct isn’t stupid at all. My guess is that what you experience is exactly the taste of change. Value changes and it makes us really uncomfortable. Even before the markets prices this in, you’re already wrapping your head around that. What is that thing, that once was distinguished and grand, still worth? And if only looking at the output, you might well conclude that is should be close to zero. Or close to time spent.
Time spent, which we diverted from for decades, cause it wasn’t a good indicator on quality of output, peeps it head again and presents itself as a worth considering, again.
Then again. It is not that effort disappeared; it moved upstream and got harder to see. Deciding what to build, framing it correctly, knowing what good looks like before you see it becomes a skill in itself and the demanding part. Often refered to as imagination, creativity, foresight .. which doesn’t really help when you just want to measure something. It’s invisible, silent, and looks a lot like doing nothing. The distinction that has to hold is between asking how did you approach this as a craft question and asking it as a control question.
So what actually gets valued, when the artefact proves nothing and the hours prove less? Probably the thing that was always doing the work and never got measured: judgment about what’s worth making at all.