Every choice that you make optimizes for something.
In school, you optimize for your grades. On social media, you probably optimize for likes, retweets, and engagement. At an established company, you might optimize for quarterly profits.
Whenever you push in one direction – “study harder,” “post a spicy take,” “cut costs” – you’re implicitly following a rule that says one outcome is good, and another one is not.
Your objective function is that rule you follow. It’s the metric that determines whether you’re winning or losing at whatever game you’re playing. The problem is that we often choose the wrong objective function, or worse, we inherit an objective function from someone else without questioning its validity.
Goodhart’s Law
There’s a funny principle called Goodhart’s Law that states: “When a measure becomes a target, it ceases to be a good measure.”
In other words, once people know what metric they’re being judged on, they optimize for that metric, even if doing so undermines the very thing that the metric was supposed to measure.
Educators teach to a standardized test, rather than teaching kids to problem-solve or learn effectively.
Banks structure their balance sheets to pass Federal Reserve stress tests, rather than developing genuinely resilient risk management systems.
The scary part is that often, these misaligned approaches stem from a well-intentioned objective function. Good-hearted people pick clear, well-meaning metrics, but they still end up optimizing for a metric instead of the actual goal.
What kind of reasoning could have led to those flawed objective functions?
- Education
- Logic: The country wants to improve nationwide education and even out state-by-state disparity
- The metric chosen (incorrectly): Standardized test scores across all schools and districts
- The resulting objective function: Schools maximize for average standardized test score
- Banking Regulation
- Logic: Regulators want to protect the economy and ensure banks stay solvent
- The metric chosen (incorrectly): Pass/fail rates on bank stress tests
- The resulting objective function: Banks structure financial holdings to pass specific testing scenarios
Picking the wrong metric to track derails the whole exercise.
This is what picking the wrong objective function can do - you’re forced into actions that aren’t actually in line with what you wanted to accomplish. An outcome that only looks like it’s progressing.
Even worse, these actions look similar enough to the end goal that you can be led astray for a long time before you notice.
MBAs
I saw this play out in real time in business school. In theory, business school is where the most promising young adults go to maximize their career potential. But unfortunately, the vast majority of my classmates flooded into investment banking, consulting, and Big Tech, maximizing for the safety of the well-trodden path.
- Logic: “I am spending significant time and money on business school, so I need to guarantee a strong financial (and socially-credible) return on this investment”
- The metric chosen (incorrectly): Starting salary and prestige
- Objective function: Maximize immediate compensation of first post-MBA role
The wrong objective function leads you down the wrong path. How did these students get convinced to optimize for the wrong metric? Well, here’s a picture from Wharton’s 2025 Class Profile.
Finding Better Objective Functions
At its core, this is all about incentive alignment. When you choose an objective function, you’re setting up the incentive structure that guides your decision-making. If your objective function is misaligned with what you actually value, your incentives will push you in the wrong direction.
The venture capitalist who optimizes for 0% company failure rate creates a system where founders are incentivized by said VC to play it safe. The grad student who optimizes for the safest path creates a system where their own talents and creativity are underutilized.
So to find the right objective function, ask yourself what actually matters to you in the domain you’re solving for. The challenge is that the things that truly matter are harder to measure than the proxies we typically use.
But just because something is hard to measure doesn’t mean it’s not worth optimizing for. If you’re not careful, you’ll spend years optimizing for the wrong game, then wonder why it doesn’t feel like you won anything at all.