See if these scenarios sound familiar:
A well-paid executive feels trapped in her respected but unsatisfying job.
A start-up founder realizes he must “fire” his paying customers so he can successfully pivot the business in a different direction.
A government planner has a new president (have you seen the headlines?) who insists on challenging outdated infrastructure and other revenue-hogging legacies.
All of these people are stuck in a “local maximum.” Yes, they’ve reached a peak. But it’s not the one that fulfills the highest potential.
Judah Taub offers some thought-provoking approaches to these conundrums in his new book How to Move Up When the Only Way Is Down.
Taub is managing partner and co-founder of Hetz Ventures. He previously was head of data at Lansdowne Partners, a $20 billion London-based hedge fund, and advised multiple young start-ups. In 2020 he was selected as one of the Forbes 30 Under 30. He served as a classified intelligence officer in the Israel Defense Forces. He lectures on time management and creative thinking at the Wharton Business School and other venues, and serves on the boards of several AI-focused companies.
What’s the origin of the name and concept of local maximum?
“Computer scientists and mathematicians spend large portions of their time avoiding local minima—points that appear to be the lowest value but aren’t globally optimal,” Taub says. “Large companies like Amazon, Google, and Netflix invest billions each year to sidestep these pitfalls because better optimization—whether in delivery routes, recommendation engines, or search results—directly impacts their bottom line.”
Taub says that while algorithms focus on minimizing, humans are often maximizing. “We aim for the ‘highest peak” in our careers or personal goals, but sometimes that peak isn’t truly the highest. We may get stuck at a local maximum, a point that feels like success but isn’t optimal in the broader landscape. Backtracking from this position to reach a higher goal can be costly and challenging.
Taub’s book is about learning techniques, many of which are derived from the way humans have trained artificial intelligence to overcome local minima. These techniques help people make better decisions, avoid the pain of local maximum backtracking, and manage the pain in the case that they must move down to access a higher peak.
What common factors may cause some people or organizations to fall into local maximum traps?
“There are numerous characteristics that make people or organizations far more likely to run into a local maximum,” Taub says. “A key one is when individuals or companies focus primarily on muscle versus agility. In this case, we can say ‘muscle’ means large, fixed investments that are very good for solving a very specific challenge but are hard or nearly impossible to adjust if the goal or requirements change. An example of this for an organization could be large CapEx investments or equipment, in highly specific factories. For humans this could be extreme specialization in a niche field or skill set.”
Thinking of it more literally, he says, an extremely muscular body trained for a specific task will outperform the agile athlete, but if the field requirements change, the agile athlete will adapt and outperform. “In the world of software and large language models, we continuously see the broader, more flexible, agile models outperform the multi-layered, ‘muscular’ ones.”
Taub says perhaps the concept is illustrated best by the now-famous quote from former Blockbuster CEO, Jim Keyes, not long before the company went bust: “I’ve been frankly confused by this fascination that everybody has with Netflix … Netflix doesn’t really have or do anything that we can’t or don’t already do ourselves.” The key part, Taub says, “is that Blockbuster did indeed have more—heavy investments in real estate, stores, DVD purchases, and other limiting ‘muscle.’ But that all translated into a more challenging local maximum when the time came to be agile in the face of major industry change.”
Some people are wary of using AI, but Taub says it’s superior to traditional methods of making good decisions.
“Many years ago, the best chess players spent their training time studying from one another and reading books,” he says. “Today, the standard for every professional chess player is to use sophisticated software to provide suggested moves in complex positions. Should the chess player always take the software’s suggestions? Probably not. However, not taking the tech’s advice occasionally would be foolish. So, too, with complex local maximums. We would be unwise to not even ask what AI would suggest.”
In what ways has Taub’s military experience affected his decision-making orientation, and what does it have in common with start-up entrepreneurship?
“There’s something unique about both of these fields—military and entrepreneurship— where in both cases you are trained to make decisions with imperfect and sometimes very little information,” he says. “And in both cases, your decision-making is absolutely critical to the outcomes. Specifically in the military, many training drills purposely start with one set of rules which would change throughout the exercise. This was to test how cadets adjust and see beyond short-term or fixed paths. It reminds me of what Carl Von Clausewitz said: ‘War is the realm of uncertainty; three quarters of the factors on which action in war is based are wrapped in a fog of greater or lesser uncertainty.’”
Taub says another way people learn about overcoming local maximums is from an exercise he participated in during special forces tryouts. Cadets would undergo a “sandbag drill” in the desert to test teamwork. Standing in a circle, they fill 20-kilo sandbags for an hour, earning points both for sandbags placed in the center (group score) and those placed individually by each cadet. “Initially, many focus on maximizing their own scores,” Taub says. “But as the drill repeats, cadets begin to collaborate. Those who work together tend to score higher and push harder, while those who go it alone end up limiting their potential. Those wide-eyed enough to notice the local maximum have a better time avoiding it, and a better chance of making the cut.”
How can startup entrepreneurs identify and overcome local maximum blocks and biases?
As a venture capitalist, Taub says, he often speaks to start-up founders at the end of the day when they’re finally finished with the day-to-day business nuts and bolts. “This is the time of day when they’re thinking of the big picture, considering the strategic question: Are we climbing the optimal mountain? While conducting these conversations regularly, one sees how sometimes climbing the right mountain can be a more important factor than nearly anything else.
Here are some of the methods Taub suggests:
- A/B/X testing rather than simple A/B testing. Specifically, this is about what you do with your research budget rather than just focusing on development. Focusing on testing A vs B to rule out a specific feature is fine; but adding an ‘X’ to the testing—a jump, a leap, a totally new factor—might just give you a whole new path you wouldn’t have seen otherwise.
- Avoiding the psychological games that can lead to debilitating obstacles. How committed are you to your beliefs, and what steps are you willing to take to unpack the validity of them? Obstacles like this often lead to founders either seeing mountains where they don’t exist, or seeing smaller mountains when the opportunity is actually enormous.
- Understanding and overcoming valleys. There are numerous techniques to use to get from one mountain to another and overcome the pain of backtracking, but it’s also critical to understand that sometimes the challenge is to shorten the width or depth of the valley between the peaks. Learning how much greener or higher the other mountain is, is highly likely to affect your ability to overcome a valley.