How DFS, poker, basketball, Starcraft and strong opinions, weakly held (SOWH) can help your startup

Since my co-founder Ricky introduced me to the concept of “strong opinions, weakly held”¹, I’ve been thinking a lot about why this concept seemed to resonate with me and how to deliberately practice it. For brevity’s sake, I’m going to henceforth refer to this concept as SOWH (pronounced SOW as in you reap what you sow).

For many worthwhile endeavors, whether that means figuring what to study in school, what to build first for your startup, or how to find the right partner, you typically start with a very limited set of information. It’s rarely clear at the outset what the best course of action is, or even what the first step should be. In school, we were encouraged to withhold judgment, keep an open mind, and try not to decide on anything until we had gathered enough information. Moreover, in group decision-making settings, when we’re aware that nobody has all of the information, we tend to try to defer to consensus building. However, when the amount of available information is so lacking to the point of feeling debilitating, we often choose to punt the decision—to make a non-decision. We think: why make a choice now when you can make one later with more information? What if we make the wrong choice? Worse yet, what if we make a choice that limits our optionality or completely shuts the door on an opportunity? It is much safer to try to gather information before committing to any course of action. However, SOWH would argue that action based on a clear hypothesis is the most efficient and effective way to get information, so long as you are willing to let go of your initial hypothesis in light of evidence to the contrary.

So WTF is SOWH and SO-WHAT? Let’s decompose and start with “Strong Opinions”.

Strong Opinions

Having a strong opinion² means that you look at the information given to you and formulate your best hypothesis h₁. You then proceed to act upon h₁ despite uncertainty. Moreover, you act as if your level of confidence c in your hypothesis h₁ is greater than your true confidence level ᴦ to start to gather evidence to increase ᴦ. To be clear, in practicing SOWH, you must first acknowledge that uncertainty does indeed exist, and also that it is generally useful to communicate your level of uncertainty or confidence in your hypotheses. Nevertheless, for many endeavors, there simply isn’t enough information for anyone to form any kind of hypothesis in which they have more than a modicum of confidence. If every party involved in the decision-making process expressed their opinions without any conviction, you would often end up in a stalemate without any clear path forward.

To move forward, you try to quickly gather all of the evidence in a single direction, then “pound the table” with your opinion so that 1) you invite everyone to come up with all the reasons why h₁ may not be correct or may not work, and 2) you can quickly devise experiments to try to prove that h₁ is not correct. The central thesis of SOWH is that action is usually the best way to gain more information, and action based upon a strong opinion brings about the best information.

So what about the “weakly held” part?

Weakly Held

While gathering evidence in support of your hypothesis h₁, it can be tricky to collect and judge evidence objectively— after all, we’re only human, and we tend to prefer to see evidence that confirms our views³. But the trick here is that we’re not looking for evidence to necessarily confirm h₁— in most cases, proving h₁ right is a long process that requires lots of evidence-gathering. What we’re really looking for is evidence to reject h₁ so that we can backtrack and formulate a new hypothesis with additional data. Backtracking a bit, let’s define what we mean by backtracking in the context of SOWH.

Let’s go back to the “what to study in school” example, and say that you’re a college freshman and you aren’t really sure what you want to study. You like math, you like physics, and you like computers. You have an inkling that it might be cool to be a theoretical physicist, but you aren’t sure what that encompasses. You could just take a broad range of classes to get a general feel of what you like or don’t like. Or by practicing SOWH, you proceed with the assumption that you’re going to study theoretical physics until proven wrong, so you take an introductory physics seminar, take the prerequisites for a theoretical physics class, or try to do research with a professor in that department. After a few classes, you may find that you actually can’t stand, well, how theoretical it all is. Now you’ve explored and ruled out one possibility definitively, and you can backtrack, but that doesn’t mean you have to start from scratch. You really enjoyed some aspects of those physics classes, especially running the experiments involving computer simulations, so maybe your next hypothesis is that you want to do experimental physics.

What we’d look like if Sheldon Cooper explained physics to us Source: Gyfcat

If you were to walk into an introductory computer science class, you would probably hear the terms “breadth-first search” (BFS) and “depth-first search” (DFS). They’re both graph traversal algorithms. In the former, you search all neighboring nodes at the same depth before moving onto the next level, whereas in the latter, you traverse down a single branch until you either find what you were looking for or are forced to backtrack because your current branch cannot possibly have what you were looking for.

First: Breadth-first search and second: Depth-first search Source: Wikipedia/Wikimedia Commons

Think of your hypothesis h₁ as the first branch you traverse in the graph—you might get lucky and find what you were looking for without exploring any other branches. But more likely, at some point you’ll hit a dead end and need to backtrack. As soon as we see enough evidence that contradicts h₁, we need to backtrack. Sometimes you’re able to immediately prove your hypothesis wrong, and other times you may traverse down what seems like a really promising path only to have it turn out to be a dead end. Sometimes, you only need to backtrack a little bit, and other times you may need to backtrack all the way back to the root and start all over. Practicing the weakly held part of SOWH means that you’re just treating truth-finding as a graph algorithm— algorithms don’t feel like their egos are hurt when they need to backtrack⁴.

When you search breadth-first, you may gather lots of evidence, but it’s likely the evidence will be orthogonal or worse yet, contradictory. Let’s say you go out to talk to users for your startup’s product. You decide to talk to five users matching very different user profiles for 20 minutes each. You’ll likely find that they all have different use cases, get different value from your product, and therefore want different features. Or you could spend a few hours with a single power user, build a prototype for her to try solve her problem really, really well, get more feedback from her, and then continue to iterate on solving her biggest problem. Now, it may turn out that the job that user wants to hire your product for isn’t a good fit, or that there aren’t many other users like her. But until you find that evidence, you proceed assuming that that user’s needs represent exactly the power users you want to build for. The key insight here is that as soon as we accumulate enough evidence that a path is not going to yield the correct solution, we must immediately backtrack. If we stay attached to our initial strong opinion or continue looking for evidence to prove h₁ even though there’s a lot more evidence to reject h₁, we’re not practicing SOWH, we’re just setting ourselves up to be wrong more often than not.

That all sounds great in theory, but in the real world, unlike a true graph search, we rarely have the luxury of collecting enough evidence to show with 100% certainty that we should reject h₁ and that we need to backtrack, so how do we know when to revise our initial strong opinion? And is this better than just exploring breadth-first or gathering evidence differently before formulating our initial hypothesis? Why do we need to form a strong opinion at all? To answer those questions, I’m going to explore three examples from three seemingly very different games: poker, basketball, and Starcraft.

Poker: The perfect game of imperfect information

Poker presents a perfect microcosm for practicing SOWH. In No-Limit Texas Hold’em, you start with seemingly very little information— just the two cards in front of you. As the hand goes on, you gain more information in the form of the house cards being revealed as well as other players’ actions. Each time it’s your turn, you have a limited set of options: you can check/call, bet/raise, or fold. If every player played with their cards face-up, the game wouldn’t be very interesting. But by hiding your opponents’ cards, we create a perfect scenario to think about how to act based on imperfect information. Even the best players in the world know they won’t be right 100% of the time, nor is that the point of the game. Playing winning poker over the long run means maintaining a positive expected value— that is, making the correct decision more often than your opponents, and assuming you avoid busting your bankroll long enough for luck/variance to even out.

A novice poker player might focus just on the information they seem to have and try to think in a vacuum: “If I have a 10♡2♢ and the house cards are J♠J♣10♠A♠, I have two pairs, Jacks and 10s, and that’s pretty good”. But that’s missing the entire point of the game— it’s not about going over the information you’re already sure of (although you should check your hole cards if you’ve had a few 🍻). It’s about using the information you already have, forming opinions about what the other players may have, and taking actions based on your opinion, while gaining more information in the process. Even before the cards are dealt, you already have a litany of information: each player’s stack sizes, their playing styles and tendencies thus far, and everyone’s position in the hand.

Here’s a more concrete example of how SOWH might apply at the poker table. You’re dealt 9♠Q♠ in the big blind of a tournament and everyone folds around to the small blind who calls you. You check, and the flop comes out 6♡7♡10♠. Given your limited information, since you don’t have a pair, and you don’t know whether you’re ahead or behind your opponent here, your default instinct might just be to check and see the next card. After all, why not get free information? However, if you formed the opinion that based on your opponent’s actions pre-flop and their playing style that they might have called with a small pocket pair or 2 high cards or that they would only call you here with a small range of hands and will otherwise fold, then you should try to test that hypothesis by betting. If they raise you, then chances are that you were just flat out wrong and they actually had a stronger hand than you thought and it’s likely an easy fold. Here’s the kicker though: if they fold, then you might have been right, or you might have been wrong, but it doesn’t matter because you won the hand. This is called fold equity in poker. Specifically:

Pot equity = Your percentage chance of winning the current pot/hand

Fold equity = likelihood that the opponent folds * gain in equity if the opponent folds

Your expected value from a hand is equal to your pot equity plus your fold equity. In this case, by acting, you not only gain more information, you’re creating opportunities for your opponent, who also has imperfect information, to make mistakes. Or your opponent might come back and immediately re-raise all-in, which should cause you to quickly backtrack on whatever you thought and reconsider all the information you’ve seen thus far in light of this new information and come to a new conclusion.

To be clear, you’re also assuming that your opponent is playing reasonably as your initial hypothesis. If they tend to just call with weak hands or decide they want to push all-in every time with 2-7 because Eddie George was their favorite player and he wore 27, once you see evidence of that, you have to quickly invalidate your original hypothesis that they’re a solid player. In the long run, if they do play like this and you are aware of that, the odds will work against them. Taking a strong opinion or betting aggressively is somewhat suboptimal, as in you’re giving away some edge in a perfect world where everyone plays perfectly. But you don’t live in a world where everyone plays perfectly. By taking action, you’re creating opportunities for you to either win outright or actually gain more information quickly.

Basketball: James Harden’s Strong Opinion

Speaking of strong opinions, I have had a strong opinion about James Harden— I was not a fan of his playing style. Don’t get me wrong— I’m still not a Harden apologist, but I have come to appreciate his greatness. For the sake of this essay, perhaps no player better embodies SOWH than Harden. He is generating offense at historically elite rates while mostly working in isolation plays. What does this have anything to do with SOWH?

Let’s treat basketball as a game of information— if the defense knew exactly what the offense and individual players with the ball were going to do, they would have a much higher likelihood of preventing them from scoring. The more ways and spots on a court a team can score baskets, the harder they are to defend. The defense gains an advantage when the offense has limited options— whether that’s because it’s towards the end of the shot clock/quarter or because of the abilities of personnel on the court, e.g. several of their players won’t shoot from outside the 3-point line. If your team comprised of Michael Jordan, two of me and two of you with all of our arms taped behind our backs, the other team would know with 100% certainty that MJ wasn’t going to pass to either of either of us⁵, so they could have 5 guys surround him, and even MJ likely wouldn’t be very successful most possessions. Therefore, the offense keeps the defense guessing by moving the ball around, attacking and forcing them to react. Traditionally, isolation plays— plays where you give the ball to one player then put the other 4 players on the other side of the court to allow that player to go one-on-one against his defender, aren’t very successful because even if that player has the advantage against his or her defender, that’s generally not as large of an advantage as the offensive team has against the defense team.

James Harden breaks all of this conventional wisdom. He scores VERY EFFICIENTLY in isolation. Harden has two strong opinions about what he’d like to do with the basketball— he’s going to drive all the way to the basket, or he’s going to create a 3-point shot for himself. Once he gets in the lane with the ball, he PICKS UP his dribble. Normally, this is a very suboptimal move because it removes optionality for the offensive player. Offensive players assume the “triple-threat” position, which means they can dribble the basketball and move around the court, pass the ball, or shoot the ball. But once a player starts dribbling, then terminates their dribble, they can only take two additional steps to shoot the ball or pass it.

Harden gives up optionality and perhaps even some optimality, but by doing this, he forces his defender into making a decision. Most defenders usually don’t gamble to go for the steal while a player is dribbling because they know that it’s risky and by committing to go for the steal, the ballhandler will just change directions and go right past them. But once a ballhandler picks up the dribble, the defender thinks they have the advantage and can now go for the steal. Harden’s defenders do often get baited into going for the steal or block, which is why Harden can draw fouls at a historically elite rate. Even if he had decided to go right to shoot a layup and would have gotten blocked by another defender coming from the weak side, beacuse his primary defender reached and fouled him, Harden wins the possession regardless of whether he was right or wrong. This is like the aforementioned concept of fold equity in basketball— by acting strongly and pushing the decision to his opponent, he can sometimes capitalize by drawing an immediate win: foul equity. As an 88% free throw shooter, getting fouled on a shot basically means 2 (or 3) free points for his team.

What about the “weakly held” part? Even though Harden has a strong opinion that he’s going to score (two-thirds of his iso possessions end with him taking the shot), he holds onto that opinion weakly and is an excellent passer, currently tied for 7th in the league at 7.5 assists per game. When the defense collapses on Harden’s drives, he is more than happy to backtrack and find a teammate for an open 3 or dunk.

Starcraft: Well?! Make up your mind

Screenshot from Youtube video by Kebooooo: Starcraft - Terran Mission 5: Revolution

Some of you may recognize this— this is from the 5th Terran Single Player Campaign in Starcraft 1, circa 1998. SOWHAT is it doing here? Let me (attempt to) explain.

We’re dropped into this map, mostly full of unknowns with your small 10-unit army. We didn’t pay attention to the briefing, so we set out to explore the map. Lots of unknowns, especially when we look at how daunting the unexplored areas of the map are:

Screenshot from Youtube video by Kebooooo: Starcraft - Terran Mission 5: Revolution

Absent any information, you might choose to explore outwards from the known parts of the map slowly. If you go back to the briefing, you remember that one of the objectives is to bring Kerrigan to the Antigan command center, so you know you should focus on doing that rather than trying futilely to destroy all of the enemies’ units with just your initial 10-person army. Of course, once you do achieve that objective, you gain a new piece of information: you actually gain control of the base, which gives you more options because you can now build more than just the units you started the mission with. However, your attention is limited and you’re now dividing it between building SCVs/buildings/units, defending your base, and continuing to explore the map to find the enemy. You eschew additional exploration in favor of building up your base and growing your army. You’ve mostly used marines in all of the levels up until now and you’re comfortable with them, so you max out a large army of marines and now you’re ready to defeat any possible base the enemy might have… except it turns out that the enemy in this level is actually on an island that your marines can’t get to without dropships.

If you instead play this level with the strong opinion that you don’t need much of an army to defeat the enemy until proven otherwise, even after you take the base, you would continue exploring/attacking where it seems like the enemy units are coming from and quickly discover that they’re on an island on the southern part of the map. You may lose some units or reach a dead end walking through the land area of the map, but you’d quickly gain the information you need to build the right units— namely that you want to build dropships and wraiths.

Playing human opponents in a 1v1 or team game made things even more interesting— rather than being dropped onto a map with lots of unknowns, you know that both you and your opponent start with just your Command Center/Nexus/Hatchery and four SCVs/Probes/Drones.

You have a general idea of early build orders before you get an idea of what your opponent is doing, and you likely have an initial strategy based on the map, your opponent’s race selection, and what you know your opponent likes to do. However, as soon as you see evidence that your opponent is expanding and spread too thin, so you decide to go for a drop on their main base. If however, you soon discover that they put up lots of air defense at the main base, you need to immediately abandon that plan and attack their expansion instead. If you try to “stick to the plan”, good luck with your vultures and tanks against carriers. The plan was just your h₁ and you need to adjust and react to your opponent as the game progresses. The key insight from Starcraft is that you need to be constantly testing and adjusting your hypothesis by scouting your opponent, all while continually executing whatever the best strategy you know of at any given point in time.

And like the poker and basketball examples above, often just by biasing for action, you may gain immediate wins regardless of whether you were right or wrong. For example, you can send your units to attack the opponent’s base at a certain point in the game, and even if your units get there and you quickly see that your attack is unlikely to succeed and immediately retreat (backtrack!), you might still gain an advantage even though you were wrong about being able to break their defense. In this case, you may gain information about the enemies’ hidden units or perhaps their hidden tech units, or you might cause them to overreact and waste precious resources on static defense.

SOWHAT?

While forming and acting upon strong opinions may seem contradictory to withholding judgment, if we practice holding onto those opinions weakly, we’re enabling ourselves to quickly get more information and continue to process information with an open mind. Being able to separate yourself from your opinion—that is, being able to separate your ego from the merit of your opinion, is essential to this process. Moreover, as we’ve seen in the poker, basketball, and Starcraft examples above, just taking action sometimes leads to outright immediate wins regardless of whether you were technically right or wrong about your hypothesis.

So why don’t people practice SOWH more often? Practicing SOWH is not easy. It involves three difficult steps:

1. Form a strong opinion based on evidence, argue for or act upon it with more confidence than you really have, and then be willing to be wrong very publicly.
2. Not tying your ego that this strong opinion which you have now put out there.
3. Quickly and readily admitting that you were wrong, backtracking based on new evidence and quickly forming a new strong opinion and repeating it all over again.

Our egos don’t want other people knowing we were wrong, or worse off, inconsistent. Nobody wants to be a flip-flopper. We view politicians who flip-flop as untrustworthy— if you can’t stand by your views consistently, then nothing you say or promise has much merit. The harder we “pound the table”, the more likely we are to get attached to that opinion, which makes it harder to let it go. We naturally get attached to our initial opinions, and because of confirmation bias, we often will see or interpret evidence as supporting it rather than looking equally for evidence for and against our hypothesis. Our newsfeed-based consumption today makes it even easier to just find things that reaffirm our world view rather than looking for evidence to the contrary. That becomes very, very dangerous. If you’ve already invested all this energy into believing in something and espousing it, you don’t want to waste all that time, energy and social capital by choosing to find evidence that you’re wrong.

But my strong opinion is that despite all of these things, it’s worth practicing SOWH in many situations, and it’d be even more costly to not. The head-fake⁶ here is that writing this post was my attempt to practice SOWH. If I’m wrong about anything here, I hope that you won’t hesitate to correct me, although I won’t say what percent confident c I am that SOWH is the correct decision framework for startups or building products 😀. But, please have at it. Please question everything I’ve presented here— my goals in writing this were 1) to practice those atrophied writing muscles, and 2) to learn more about this way of thinking/acting, especially if you have better examples. If there’s anything I should read related to this topic, I’d love to know as well. I’ve been working on another essay about a related topic— how startups should optimize around the optimal. Follow me on Medium, subscribe via email/RSS, or check back here shortly.

In the words of Edmund Duke, “Well?! Make up your mind…” and the Terran vulture: “Somethin’ on yo’ mind?” Leave a comment below.

Thanks to Ricky Yean, Shravan Reddy, and Sidney Le for reading drafts of this.

Further Reading on SOWH

• Paul Saffo
• Jeff Atwood
• Ricky Yean
• Ameet Ranadive

Footnotes