What do you really care about?

Recently, I read that some software companies - big and small - are creating leaderboards for their employees based on how many AI tokens they use.

As modern humans, especially software engineers are trained to optimize for all kinds of metrics and KPIs, it is obvious what such leaderboards produce: They increase the usage of AI tools in these companies. A process called tokenmaxxing.

But why do they care? Well, it is possible that there is a smart plan behind it: Maybe some of these companies are or want to be in the business of selling AI solutions, and are just sourcing ideas unbound by resource constraints from their employees. Maybe there is a system behind it that categorizes all the outputs and evaluates them for a potential business use-case.

Another possibility that I can imagine: Companies with AI leaderboards really just want to unleash their employees creativity for a while, and have some other accountability metrics running in the background to assess this experiment with regard to their mission, based on which lessons can be learned.

Unfortunately, for many or maybe all of the companies with tokenmaxxing leaderboards, this might not be the case.

So where does the impulse really come from, if not from a clear plan?

Especially when the market becomes as uncertain as it is right now, decisions can be driven by two emotional phenomena: lack of self-worth and fear.

And from the outside it looks more likely to me: If a leader or a company doesn’t have a mission they believe fits in the current market environment, or maybe they have no mission at all in the first place, there is nothing that answers the “Why are we doing something?” or the “Who are we doing this for?”.

And suddenly the most important question “What does a company do that is defined by our mission?” turns into something like “What do we have to do such that we don’t fall behind?” or “What do the big companies do?”.

In that case, the power for decisions that should lie inside the company is outsourced to the competition or some arbitrary market leaders. And yes, that happens all the time. Because leaders are overwhelmed, or simply don’t feel the self-worth to answer the question: “What do we actually care about?”.

Again: I don’t know if it is the case for all the companies that implemented tokenmaxxing leaderboards to assess their employees’ productivity. Probably for some companies that might make sense, maybe some companies profit from punch cards or lines of code - productivity measures, but - as I understand it - that has never been a lasting phenomenon.

Large language models certainly can produce a lot of stuff in a very short time, which is a definition for productivity, but how do we assess that we care about the stuff?

The same dynamic plays out at the personal level too. I recently had a little chat about design choices for a website project. And I felt really smart about what I learned from it. As I am not a web designer and have not given deep thought to how we can express functional goals with colors, layouts, fonts and structure, I have trouble even starting to think about how to make decisions in this regard. And the LLM certainly kick-started me here. But then: What would an actual web designer have noticed? Probably much more than me.
Does the LLM quality suffice for my current needs? I believe so, because at this point, I am still working on the functionality, and in my opinion form follows function. Do I really want to be a proficient web designer, and pay attention to it? No. So, at some point, I might want to hire a professional.

In this case — as my own metric is not tokenmaxxing — I can ask those questions freely, and answer them based on my own wisdom. But that requires knowing what I actually care about.

For enterprises that know what they stand for, blindly relying on a metric that changes their employees’ behavior is not the answer. After all some of the most successful engineering businesses rely on human noticing as much as possible, such as Rolls-Royce or Rolex, which rely on humans to do jobs that could be automated, such that possible improvements can be noticed.

Decision-making based on a lack of self-worth and fear is a phenomenon that is not exclusive to business. If we do not trust our own wisdom with regard to our own health, our relationships, childbearing, child-rearing, if and where to go on vacation, or any aspect of our life really, we will rely on outside influences, such as Instagram advertisement algorithms for bringing us the answers.

Reflection of the Day

• If you are a company leader, make yourself aware of a metric by which you assess your team’s productivity. Can you explain how it serves the company’s mission?
• In your own life: Do you have a set of values or a mission that guides you? Write them down.

With care,

Martin