The Fox and the Hedgehog: Will Tom Thibodeau adjust his approach in future years?
Tom Thibodeau is the NBA’s Coach of the Year after leading the Knicks to an identity-defining 41-win campaign this season. But will Thibs be able to adjust his approach in future years to avoid losses to teams like the Hawks again?
As we draw the curtain on the magical 2020-2021 New York Knicks, it is now time to look forward to a bright but still uncertain future. Tom Thibodeau’s achievement to bring the Knicks from a laughingstock to 41 wins and a fourth seed — officially cementing himself as the well-deserved NBA Coach of the Year — cannot be understated. Whatever happens, we’ll always remember this magical season.
But the playoff loss brought back some of the criticisms we’ve heard leveled at Thibodeau, most notably him being stubborn and not making offensive adjustments in the playoffs. Jason Concepcion, AKA netw3rk, has drawn the ire of his fellow Knicks fans at times, but I thought this was a pretty succinct outline of the trade-offs (and why at this moment in time, they are trade-offs we will all gladly make):
All of this is true. There were changes (going small, balancing creation and shooting between the lineups better, defensive adjustments) that we were all screaming for Thibs to make, and yet it likely would not have made much of a difference against a team overflowing with shot creation, getting healthy at the right time, and having an elite, mobile rim protector.
But the part of his tweet I found interesting was about Thibs having a “sell-by date.” Many have made the case that Thibodeau is excellent for developing a base level of competence, effort, and execution (particularly on defense), but isn’t the guy to get you to the level of true title contention. This is the question I’d like to focus on in this piece, but before that, I’d like to outline a framework from which to judge this: the fox and the hedgehog.
There is an ancient fable (included in Aesop’s fables) about The Fox and the Cat. A fox and cat are conversing about strategies to avoid predators. The fox, clever as one might expect, claims to have many varied strategies to avoid capture. The cat admits that it has only one. The duo is then approached by hunters, at which point the cat immediately climbs a nearby tree. The fox thinks of his many strategies and which one would be best to deploy, but in that time he is captured by the hunters’ hounds.
Does that sound familiar? I don’t want to turn this into yet another David Fizdale criticism piece, but it was notable how often the Knicks seemed confused and lost on both sides of the floor under him. Without a set defensive strategy (and with many young players), the Knicks were plagued by miscommunication and blown coverages. In fact, when Mike Miller took over, one of the lowest-hanging fruits he was able to grab was implementing a simple, repeatable scheme the Knicks could learn and develop competence with. It wasn’t always the most effective against every attack, but it limited mistakes and put the onus on teams to beat them without the Knicks beating themselves.
A similar case can be made for the offense. Julius Randle struggled immensely in his first year as the focal point of the Knicks’ offense. While much of this can be attributed to needed improvements in ball handling, feel, and shooting, Randle himself discussed how much easier the game was made for him by Thibodeau this year. He specifically mentioned early in the year how it was easier for him to make reads because he knew where everyone would be.
Randle clearly benefited from this en route to a breakout 24/10/6 season, but the effect on the supporting cast was immense as well. RJ Barrett is a young player, so improvement is to be expected, but jumping to 40% from three (including 37% above the break, and miss me with the “empty gym” stuff) was unprecedented. Reggie Bullock had a career year after shooting just 33% from three last year. While some of this can be attributed to continuity, it’s clear that simplifying the offense, playing through Randle, and having the Knicks operate with basic offensive principles that they were able to get very comfortable with really greased the gears of a formerly stagnant offense that still lacked talent.
Now, of course, there is a tradeoff between simplicity/ease of implementation and versatility. If you get really good at climbing trees, it’s certainly better than being stuck in analysis paralysis without a plan. Of course, if a hunter shows up with an axe (or has an eagle, or brings Spider-Man with him), it would be nice to have a Plan B. And when Trae Young and Clint Capela showed up with a shorter rotation, detailed prep, and playoff-level intensity, unfortunately, the Knicks didn’t have one.
And to be clear, this goes beyond scheme. Thibodeau also tried to keep continuity in his lineups as much as possible. This aimed at the dual objectives of building cohesion and maximizing wins. Although it certainly appears a lineup with Randle and Obi Toppin could have unlocked more opportunities on offense for the Knicks against the Hawks, the two had barely played together in the regular season. Early in the season, this was due to Obi still being wet behind the ears (particularly on defense) and later in the year was due to a desire to maximize wins with a rotation that was starting to click.
A similar argument can be made for the decision to continue to start Elfrid Payton (and to a lesser extent, keep Frank Ntilikina in a “situational” role). While this writer does not think the decision to keep starting Payton positively contributed to winning, one can see how Thibodeau would have thought messing with a lineup that was winning games to get more reps for younger players like Ntilikina and Immanuel Quickley could be counterproductive and set back some of the cohesion and comfort that had been developed.
As I’ve made clear on many occasions, I think this was an incorrect deduction, but I’m not here to litigate the decision. Enough words have been spilled on that. Rather, it illuminates Thibs’ approach. The Knicks may be short on talent, they may not be able to execute against every scheme and every opponent, but dammit they are going to be great at climbing that tree and you’re going to have to chop that sucker down if you want to catch them.
The Fox and the Hedgehog
The problem is in the playoffs, everyone knows you’re going to climb the tree, and everyone has an axe, and everyone has the time to learn and prepare against every escape you have. The Fox and the Cat parable makes the case that it’s preferable to be the cat to being the fox, but is that always the case? A similar phrase in Greek states “The fox knows many things, but the hedgehog knows one big thing.” A fox has a million different ways to avoid capture, whereas the hedgehog has its spikes. It never has any doubt about deploying those spikes, and it is going to be an effective defense against the vast majority of threats. However, spikes can only do so much if the hedgehog crosses the street and gets hit by a car. In his essay, “The Fox and the Hedgehog,” Isaiah Berlin applies this framework to evaluating the mental approach of many famous philosophers and writers. Some view the world through one overarching framework (he gives the examples of Plato, Nietzsche, and Dostoyevsky, and for a modern example I’d add Nassim Nicholas Taleb and his ideas on risk), while others draw from a wide variety of experiences (Herodotus, Aristotle, Shakespeare).
He doesn’t pass judgment on which is preferable, but I’m sure we’ve all seen both phenomena anecdotally. Sometimes we see recurring patterns in everything, like Neo reading the code of the Matrix; sometimes we have to start from scratch and build novel approaches from new experiences constantly. The risk with the former is false identification of patterns (instead of seeing the same truth everywhere, you’re seeing windmills everywhere), while the risk of the latter is being stuck in analysis paralysis and not being able to progress to any meaningful place without the ability to generalize. To the man with a hammer, everything looks like a nail, but to a man with too many tools and no comprehensive understanding of when to deploy them, everything looks insurmountable and illegible.
How do we balance the two and which is preferable? I’ll note two contrasting takes here (sorry, I promise we’ll get back to basketball soon). Political scientist Philip Tetlock has dedicated much of his research to understanding the ability of humans to forecast events. He did an extensive study on forecast accuracy for political events (like the probability that the Soviet Union would fall by a certain date). He included experts, novices, and a wide range of subjects. What he found was that being an expert had little impact on the ability to make accurate forecasts; rather, the most accurate forecasters had the characteristics of what he deemed “foxes.” These foxes tend to be skeptical about grand overarching theories, more likely to couch their predictions probabilistically, and willing to change their minds when confronted with new information (in other words, NERDS!).
In contrast, what he called “hedgehogs” were more likely to try to fit events into grand theories, less likely to change their mind and incorporate new information, more likely to be extreme partisans, and as a result of this, less likely to make accurate forecasts. Again, this was independent of level of expertise. “How you think” (being a fox) matters a lot more than your level of subject matter knowledge or expertise. (For more on this, I highly encourage readers to check out Tetlock’s excellent book, Superforecasting: The Art and Science of Prediction).
Of course, making probabilistic predictions is rarely anyone’s main objective. As Mike McDermott tells Joey Knish (shoutout long-time Knicks fan Jon Turturro) in Rounders, “You see all the angles, but you don’t have the stones to play one.” Ultimately, understanding and prediction isn’t enough; in the real world, you have to act and execute.
So if prediction is the focus of political science professors, let’s get a different perspective from a business school professor. Former Stanford professor and management guru Jim Collins writes:
“Those who built the good-to-great companies were, to one degree or another, hedgehogs. They used their hedgehog nature to drive toward what we came to call a Hedgehog Concept for their companies. Those who led the comparison companies tended to be foxes, never gaining the clarifying advantage of a Hedgehog Concept, being instead scattered, diffused, and inconsistent.”
He further writes, “A Hedgehog Concept is not a goal to be the best, a strategy to be the best, an intention to be the best, a plan to be the best. It is an understanding of what you can be the best at. The distinction is absolutely crucial.” (Sounds a whole lot like something Tom Thibodeau would say!).
(For the full blog post by Collins, see here. He has also expounded on this idea in his terrific book, Good to Great).
And here we can understand how Tom Thibodeau might approach Year 1. You can be indecisive, try a bunch of different things, and never get good at anything; or you can focus on one strategy that might not be the best, but can be your bread and butter, and be formidable with that.
How long have we complained that the core problem for the Knicks was that they lacked an identity? Fizdale never seemed to get them to adopt one beyond the poorly-defined notion of being “dogs.” When Thibodeau was hired, many believed establishing this identity had to be his first, second, and third priorities; in addition — given his background and the fact that the Knicks had an extremely talented rim protector in Mitchell Robinson, but desperately lacked offensive creation — the logical starting point for that identity was defense. And of course, Thibodeau delivered on that.
Exploitation vs. Exploration
A key research area in machine learning and artificial intelligence (most notably, the area termed “reinforcement learning”) is the idea of exploitation vs. exploration. It’s often referred to as the “multi-armed bandit” problem. Imagine you are in a casino with a thousand slot machines. Each slot machine has different odds of success. There are various strategies you can try to maximize the amount of money you win. You can sit down at one and keep playing until you win, especially if you can quickly figure out the odds. Alternatively, you can try every slot machine in the place until you find the “best” (the one that gives you the best odds of winning) one. If you pick one (or only explore two or three), you risk get stuck on a bad slot machine. If you keep jumping from machine to machine too quickly, however, you run the risk of never learning much about any of them or knowing which one is the right one, even if you stumble upon it at one point.
Similarly, when an artificial intelligence algorithm has to test and learn the outcomes of different strategies (which can be costly in the short-term) while also picking the best one to maximize its reward as quickly as possible (which can be costly in the long-term if you don’t sufficiently explore the available options) it faces this exploitation vs. exploration tradeoff. After a couple of tries, it might find a strategy with decent rewards; does it keep hammering this one, or does it try to see if there’s a better one? For a real-world example, we face this dilemma every time we want to go out to eat. Do I go with the old reliable pizza place across the street, or do I try the new fusion place? I may never discover new foods that I’d love if I stick to the same spot, but if I keep trying new places I know little to nothing about, I run the risk of having a bad meal.
Of course, you need to account for both of these objectives and make the appropriate trade-offs, just as you need to be able to marry the prediction accuracy of foxes with the decisiveness and execution of hedgehogs. At a given time, you likely have to prioritize one approach over the other.
This year, it made sense for Thibs to be a hedgehog. The Knicks had nothing to hang their hat on; nothing to exploit and no baseline from which to explore. What’s the point in exploring a million strategies if you’re not actually good at any of them? They now have one slot machine they can always go to and know what they’re going to get: tough defense that keeps offensive players out of the paint at all costs, a versatile but flawed power wing on offense, and shooters.
On defense, in particular, we saw the benefits of this exploitative strategy. Early in the year, the Knicks kept teams out of the paint at all costs; Thibs wants his help defenders getting two feet in the paint. The result of this was that the Knicks gave up a ton of open threes, and many writers (notably, friend of the Strickland Seth Partnow) wrote about anticipated regression to the mean for opponents who had been missing wide open threes against the Knicks.
But a funny thing happened on the way to the regression: not only was it not quite as dramatic as many had feared, but the Knicks adapted and improved their execution of the scheme, doing a better job of both helping aggressively in the paint, but also recovering to shooters to deny the attempts (and Partnow also noted this many times). And they became adept at executing this strategy, becoming an elite defense along the way.
Now, of course, every scheme has a kryptonite. In the Knicks’ case, helping aggressively and being adept at scrambling will work against 95% of creators. But against an elite passer who can anticipate the help and even manipulate it (yes, grudgingly, I have to admit this describes Trae Young) you’re in danger. The Knicks were successful on defense against most teams because they force you to make difficult passes to shooters who must get the shot off quickly. But if you can make that pass, you will get open looks.
“Make life hell for anyone but elite point guards” is a solid hedgehog strategy, and a worthwhile one for the Knicks to implement in Year 1. You will win a lot of games that way, because most teams don’t have the combination of point guard play and shooting to really make you pay. To get to the next level, however, the Knicks will have to get foxy.
As a long-suffering Michigan football fan, I’m reminded a bit of the defensive approach implemented under Jim Harbaugh (I promise I don’t mean this as an insult to Thibodeau). Michigan was coming off a 5-7 season and were lightyears behind Ohio State. Harbaugh came in and hired defensive veteran defensive coordinator Don Brown from Boston College, where he had consistently led elite defenses according to nearly all advanced statistics. Brown’s philosophy was quite simple: blitz, blitz, and blitz, play aggressive man coverage, and make quarterbacks hit very difficult throws over the top of long cornerbacks or hit difficult timing patterns in small windows. And against most teams, this worked extremely well! Most college football teams don’t have a quarterback who can make those throws or the kinds of receivers who can generate enough separation to threaten the defense consistently, particularly deep. Playing this man coverage allowed Brown to be aggressive pressuring the quarterback and stack the box against the run; if you don’t have an elite runner at quarterback to even the numbers or a quarterback who can punish this aggressive coverage scheme, you’re going to have a bad time.
But Ohio State and elite teams do have the talent to exploit these schemes, and Michigan was exposed when they did not have a Plan B (zone coverage would be the natural counter, but again, the hedgehog strategy and lack of zone played against poorer opponents did not lend itself to on-the-fly implementation in a marquee game, not completely dissimilarly to the Knicks’ lack of ability to adjust against the Hawks). Michigan was never able to catch Ohio State, and while the team got back to a level of respectability (like the Knicks have), they have since plateaued, and Brown was let go this past offseason.
OK, so we know the trade-offs Thibodeau made, we know the benefits we were able to reap in the regular season, and we know some of the flexibility and ability to adjust his schemes may have cost the Knicks in the playoffs. I was much more sanguine about this after Game 4, but in Game 5, when it became clear early on (even when the game was close) that the Knicks’ offensive approach was not going to change, and we were not going to see Obi and Randle together no matter what, I felt myself getting annoyed. Eventually, we’ll need to have those other tools in our bag.
But just because Thibodeau played it this way this year (and as I’ve tried to make the case above, justifiably so), does that mean he will play this way every year? The truth is, we don’t know. We know Thibodeau is a hedgehog in Year 1. But we don’t know what Thibodeau will look like in Year 3 of a team with a developed RJ Barrett and Immanuel Quickley, a more seasoned Julius Randle, and a roster with upgraded talent (and perhaps another star). With a roster that has achieved a baseline level of competence, cohesion, and continuity, you can start to experiment more and try new things. With a roster that has the talent to make the consequences of bad lineup decisions here and there more forgiving (and the ability to win enough games where you can afford to lose a couple due to experimentation without losing playoff ground), who’s to say Thibodeau won’t have an eye on making the team more dangerous in the playoffs, even if it costs you a couple of wins in the short-term?
The truth is, we don’t know. Right now, Year 3 Thibodeau is Schrodinger’s Thibodeau. The Chicago years are not as helpful given the sheer number of injuries the team suffered in Thibodeau’s later years, and Minnesota obviously never got to that point. We get to choose what we believe. Personally, I think Thibodeau correctly diagnosed the Knicks’ need to develop a core identity, versatility be damned, and build from there. He has emphasized the importance of learning and evolving (discussing embracing corner threes and playing faster on offense in press conferences is encouraging) and I believe he will continue to do so. There may come a day when we remark on just how foxy Thibodeau has become. Or maybe not, and Concepcion’s remark about the sell-by date will be proven to be prescient. All we can say, for now, is Thibodeau took the best available option to him given the circumstances; how he evolves will determine how much more the Knicks can become.
