Being Unpredictable Is a Bad Strategy | Hacker News Reader

Being Unpredictable Is a Bad Strategy

hbr.org

73 pointsapress43 comments

In chess, amateurs play for tricks, masters play for position.

Amateurs assume competitors will fall for the tricks, but that is dangerous in business. Competitors could be sandbagging, and there's no ELO based on many prior games. You're also threatening their business and livelihood, not just the loss of a 2 hour game.

Masters assume competitors can see all the tricks, so they play for better pawn structures and piece placements. It is grinding out slight long-term advantages and converting it to a win in the end game. Grinding is the only way, because the players are so equally matched in the short-term: their pieces have similar powers, and they are limited to one move per turn.

In software, businesses can build only so many features per month due to the mythical man-month. Businesses are building on top of more or less the same technologies and mathematical techniques. Businesses understand what is important to customers and how customers use the features. Businesses watch each other's product demos and talk with each other's customers, so nothing released is hidden for long. Businesses also poach talent from each other.

Scholar's mate won't work here, there's no silver bullet features, and anything you build can and will be copied. You'll have to grind it out, fighting for better pawn structures, bishops on long diagonals, network effects, brand, and talent. And in the long-term, you come out ahead.

In a perfect information, turn-based game like Chess, there are no secrets. You know where all the pieces are, who moves what when, and the goal each person has. Direct subterfuge is simply not possible.

Chess masters use unpredictability, but in a perfect information, turn based game, unpredictability is indistinguishable from other traits like adaptability or position. Furthermore, their "tricks" are aimed at achieving proximal goals -- capturing a pawn, splitting a bishop pair, a better position. These proximal goals are then leveraged for victory. This speaks more about chess (or other, perfect information, deterministic, turn based, head to head games) than about mastery in general.

In a game where much information is hidden and victory can be sudden and swift, optimal strategies will take absolutely take advantage of deception, aggression, and risk taking.

I'm hating on your analogy, but I agree with your message. In most games, predictable things are predictable because they are among the most beneficial and safest courses of action. Doing something unpredictable carries with it the implicit cost of doing something that is materially worse, and hoping you will come out ahead because of side effects such as your opponent's confusion. You're probably much better off focusing on executing those simple, obvious, and predictable things well while your opposition bleeds to death while attempting a never ending cycle of net-negative cunning deceptions.

I think the author here is using "unpredictable" as a term of art and not in a context that's well understood by me.

"Strategic" is not the opposite of "unpredictable" -- "predictable" is. Strategic thinking is identifying a long-term interest/objective/aim and developing the means to achieve them. The application of strategy isn't necessarily predictable to some parties at all. You can certainly choose or be forced to respond unpredictably in the context of a broader strategic aim.

Taking away the catchy title and fancy discussion and the point it pretty basic: emotional/intuitive responses are generally inferior to logic/fact driven planning.

Yes, I found the usage odd. He equates "unpredictable" with "random", which is not how I have usually heard the term in strategic contexts. That means unpredictable, even assuming perfect information. The usual meaning of "unpredictable" is more like the author's use of "secrecy", meaning "denying the ability for competitors to confidently predict the next move".

I don't know why unpredictability must always be considered a strategy all its own and not part of a larger strategy.

I've been reading a lot about John Boyd and the OODA loop, and one of the major takeaways was that if you have a "fingertip feeling" on changing battle conditions, acting unpredictably from time to time can disorient the enemy commander, expand their loop and therefore allow you to act at a faster tempo than he. This idea was not new but something he adapted from Sun Tzu's "cheng" and "chi", his two fundamental moves of war, from which all maneuvers and strategies are built.

> I've been reading a lot about John Boyd and the OODA loop,

Care to share your reading list? I've read the Coram biography and really enjoyed it, but it's understandably light on details of how to apply OODA to things like business.

Just Coram so far. There's a book called "Certain to Win" that applies his ideas to business.

Certain to Win is very good.

There's also sort of an "anthology" book on Boyd/OODA called Science, Strategy, and War by Osinga.

You can find some other random content on dnipogo.org and in the web archive for belisarius.com (I don't know why this is down now).

When reading about Boyd and his ideas, I think it's best to focus on the core concepts and reason about applications yourself. He built his philosophy bottom up (he was originally a fighter pilot). Examples like in Certain to Win are looking for parallels that exist coincidentally... or perhaps a better word is patterns ( http://www.dnipogo.org/boyd/patterns_ppt.pdf )

Usually predicability vs unpredicatability maps to zero- versus non-zero-sum games.

In a purely competitive (zero-sum) game, the equilibrium strategies are randomized ones which deny the opponent any exploitable openings. But you get much more interesting structures in non-zero-sum games, where both players have an incentive to coordinate with each other.

War and fighter combat is quintessential zero-sum games, but one would hope that business is win-win.

Also depends on 2-player vs. many-player games.

A few feints to throw off your opponent can work very well in a 2-player game, where, like the grandparent says, they throw off his OODA loop and can give you additional room to maneuver. In a multiplayer game, though, it's unlikely that all your competitors will be thrown off. One will go for the ideal strategy, irrespective of competitors, and while you & your chosen adversary are duking it out with feints and countermoves, they'll silently come up and pwn you all.

I'm reminded of the Microsoft/Netscape battle during the first browser war. The two of them set each other firmly in each others' sights, each trying to predict what the other would do and counter it. In the end, Netscape died and Microsoft became irrelevant, and it was a tiny startup called Google that ended up winning.

That depends on the timescale. While Google is winning in 2016, in 2006 we would definitely have been saying Microsoft had won. Nobody knows who will be winning in 2026.

The two lessons from this:

- Microsoft got so good and prolific at fighting on their terrain, they overspecialized and forgot how to win anywhere else. Once their opponents stopped stepping on their terrain, MS started losing.

- There's no such thing as the first browser war or the second browser war or the operating system war, and there's no such thing as winners or losers. There's just a never-ending sequence of skirmishes and battles.

Came to say this.

A head to head game in the vacuum of a single match, like, say, Chess or Go or Starcraft, is conducive to certain kinds of strategies. You have one opponent and no teammates. Your win is his loss. It's black and white, clearer what you're trying to accomplish.

Not all games (or "games", i.e. business) are this simple. There can be multiple players. You may have teammates, or you may need to forge alliances. You may then need to break alliances. There can be multiple degrees of victory or defeat, and multiple paths to victory.

Being unpredictable is inherently scary. It's effective in direct conflict because your opponent must account for more possible future situations, diluting his ability to attack you directly, or thwart your true attack directly.

Exactly what makes this effective in direct conflict, makes it a liability when seeking alliances. People will not want to cooperate with you if they aren't comfortable in their prediction that cooperation will be beneficial, so they'll only do it if the alternative is worse. Being predictable makes it easy for people to work with you. This opens you up to many opportunities for cooperation you'd otherwise miss. When cooperation is essential to victory, this is immensely powerful and puts a hard limit on how unpredictable you want to be if you'd like to win. And heck, in life and business, maybe relationships with people are most valuable trophy anyway, i.e. https://youtu.be/dQw4w9WgXcQ

> acting unpredictably from time to time can disorient the enemy commander, expand their loop and therefore allow you to act at a faster tempo than he

I disagree that this is a good idea. I think acting unpredictably gives your opponent a reason to look deeper into your strategy which ultimately might end up sinking it. If you perform an unpredictable move it should be for a hard and fast strategic gain.

Ideally you should move in a fashion that is entirely predictable by your opponent without incurring significant loss to yourself. This will caused them to be surprised when you eventually break from your predictable normal activities and perform an unexpected maneuver.

Hmm, what you're saying makes sense, because the way you explain it makes it sound like it would have the opposite effect and let them act inside your OODA loop. Maybe I'm reaching too far, since it's not a topic I'm well-versed in.

I've heard cheng and chi translated in other ways, like "ordinary" vs. "extraordinary", "direct" vs. "indirect," so maybe "unpredictable" doesn't really apply here. I read it as being more about conflict within an ever-changing environment. If the world is constantly falling out of order, the advantage doesn't go to the one with the most consistently rational strategy, but the one who responds to change the fastest. So instead of making your opponent confused about your strategy, the goal is to degrade his perception to make him unable to keep up with reality itself. This is all very theoretical and abstract though.

I always liked orthodox and unorthodox. I think there are lessons when Sun Tzu discusses cheng/chi that are somewhat orthogonal to the OODA loop, but I see the stated interpretation by porpoisemonkey as a "subset". Lulling your opponent into complacence with the obvious and baiting him into a lethal trap absolutely fits the definition of destroying their perception of reality, but isn't the only way.

Your opponent becoming so paralyzed by the fear of not knowing what you will do next and thus doing nothing also fits the OODA loop in a completely different way, but isn't really cheng/chi. This is also touched on in Art of War, so it's not like they are conflicting ideas.

Interesting article.

Not sure I totally find this persuasive: The author tries his damnedest to draw a distinction between "unpredictability" and "secrecy", but it ends up being pretty murky.

For example:

> Keeping secrets can protect competitive advantage. Imagine the D-Day invasion of Normandy if the Allies had announced dates and locations.

> But secrecy is not the same thing as unpredictability... Unpredictability bluffs, postures, and palters to gain advantage through uncertainty and misdirection.

Wasn't a big part of the D-day strategy to feed misinformation to the Germans? Weren't the beaches at Normandy significantly depleted on the German side because the Germans thought the allied invasion was happening somewhere else (Italy? I don't recall offhand).

His other points are similarly iffy:

> The leader — the strategist... They propose a race rather than a duel.

Isn't proposing a different contest a fairly unpredictable move? Secrecy and unpredictability seem more-or-less synonymous when viewed through the lens of your competitor.

It seems like by "unpredictable" he means "making moves that seem to provide no net benefit," not actually "unpredictable."

It depends in some cases you can't be unpredictable to your competition without being unpredictable to everyone else. In other cases you can. Strategy is about knowing when you can and can't be unpredictable. It's fine to be unpredictable to make your competition think that you want a company when you don't, so they waste a bunch more money than they should in acquiring it. Not so with your next product.

I don't disagree with you, but that isn't really what the article is saying. The article is all but making a blanket statement that "unpredictability = bad in any context":

> Behaving unpredictably with one group — customers, employees, competitors, suppliers, etc. — means exposing your unpredictability to all.

What kind of "unpredictable"?

Game theory says that in certain circumstances the best option is to pick randomly with calculable probability from known options. But, anyone using the same model of the situation will come up with the same options and probabilities for that choice. So does that really count as "unpredictable"?

(I also have doubts about how often things like that will actually show up in business in the real world, but even assuming that they do...)

Repeated negotiations benefit from random behavior. Each side benefits from information disparity, so if they always use identical approaches they start leaking more information than they want to.

Probably that should read, "ONLY repeated negotiations benefit from random behavior." Random choice among N possibles will select the best only 1/N of the time.

ONLY would mean all one off transactions must be deterministic strategy. I am not sure that's the case due to uncertainty.

I think the author's word choice of 'unpredictable' refers to Trump-like tactics in business and media: do they work or not?

I think the answer depends on where you are in the market -- do you deliver commodity or novelty? If you're a parts supplier or a contractor, you need to be reliable; you deliver an essential service on-time at a fair price. In that space, unpredictability is clearly a liability.

But if you deliver a product with features that must differentiate itself from the competition in the eyes of consumers, you must stand out: by hook (being better) or by crook (merely appearing better). Here unpredictability may indeed serve you well, especially if you need to hide your lack of exceptionalism.

But unpredictability can also arise due to originality, where your new/better approach is perceived as unexpected and destabilizing, inviting exploration of new capability rather than mere exploitation of old ones. But for this tactic to succeed, it requires genius -- insights into possibilities and capabilities that others have missed. Otherwise, your unpredictability isn't really originality, it's just clever marketing (crook, not hook).

> Ask Coca-Cola for its secret recipe and listen to a whole corporation laugh

Coca-Cola is a highly vertically-integrated company with huge economies of scale, giving it a near monopolistic position. Even if you had the 'secret' recipe, there is little hope of being able to profitably compete on price. And if you could compete in some tiny segment of Coca-Cola's market, they could lower prices and run at a loss in that segment until you ran out of money. This is before you take into account brand value...

Ask Berkshire Hathaway to invest in a sugary drinks company based on its secret recipe and listen to a whole corporation laugh.

Reinforcing your point, the "secret recipe" has been revealed, to no ill-effect. :)

http://newsfeed.time.com/2011/02/15/is-this-the-real-thing-c...

Of course the special arrangement to grow Cocoa in the US is a powerful barrier to entry as well!

Coca. Cocoa is for making chocolate. If they made their drink with that, I'd probably drink it.

"Damn You, Autocorrect!"

Additionally, recipes change overtime to accommodate any number of environmental factors: better tech, lower or higher cost/availability of raw materials (e.g. bad harvest, war). One benefit of keeping the recipe 'secret' is ability to continually tweak it. So, it's more a question of the recipe from when (and perhaps even where).

This is very evident in the history of beer. Here's just one small example: http://barclayperkins.blogspot.com/2013/02/tetley-porter-185...

Same beer across 38 years; changes in ABV, changes in hopping, changes in boil time.

Honestly, I'm a bit disappointed in the quality of this article, especially given it tries to tackle something that is so non-intiuitive. Unpredictability runs the universe, so it's evidently not a horrible strategy at some level, given I wrote this.

The author states unpredictability is NOT the same as secrecy, but then goes on to say "building fake factories" is not a good strategy. If a corporation built fake factories, then the fact they were fake would be a "secret" that only they knew about for a time (diversion).

On the other hand, if the corporation randomly picked an activity that was obvious, such as diving into a new product line, the new product line would be real and there would be no secret behind the move (given it was randomly chosen to be produced). The "purpose" behind the move may reveal itself as added profits by taking existing production efficiencies and applying them to a random product. Or, it could be a tactical maneuver to confuse or disrupt the competition (in other words, it could also be a diversion).

The problem with these types of discussions is that they typically fall into the bucket of "talking about talking about doing something". One will have a hard time seeing the benefits of it while only discussing it.

Being unpredictable is not the same thing as being random. In chess or even football or other sports, you don't want to telegraph to your competitor what your next move is. For example if the other team knows you always run a draw play on first down, then obviously that becomes easy to defend. Or if you always use the same opening in chess, etc. So unpredictability as a strategy can work but only if you do something that makes sense. Just being random is not a good strategy. Making it difficult for your competitor to anticipate your next move can be a good strategy. In international affairs however, being unpredictable can be a liability because if your advisory is not sure if your next move is to push the big red button or not, then they may mistakenly retaliate too quickly. I think that is what concerns a lot of people including me.

Just to be clear, the guy's talking about Corporate Strategy. And he's correct in a way yet incorrect in others. As he says, secrecy may lead to competitive advantages (e.g. Apple's control of the supply chain meaning advantages in both New Product Development and in Marketing and Sales).

But some companies take advantage of the public's perception of their unpredictability, and foster the perception, e.g. Paddy Power's marketing campaigns, e.g. Ryanair's pricing strategies. Obviously unpredictability for unpredictability's sake - i.e. nonsense manoeuvres - is senseless, but there's not necessarily anything wrong with a perception of unpredictability.

In Mixed-Strategy Nash Equilibria, the amount of unpredictability to use in alpha-beta-like competition games to maximize expected gains assuming a similarly motivated adversary.

In many cases these strategies are dominated by a fully deterministic strategy, or with the mixed-strategy that is highly uneven in its distribution: favoring that one action be taken almost all of the time.

The title taken on face value is flatly wrong but as headlines go it's trying to be cute and attention grabbing. Strategic surprise is a fundamental tool to strategy, but should never be assumed to just work without explicit motivating reasons (be they heuristic or calculated).

Unpredictability can confuse your competition and you can gain some advantage on that. But it can also confuse you own internal resources and make some real damage.

imo, the only time your business should, from the outside, look unpredictable is when others simply don't see the course you have charted. Even then-- markets are pretty easily segmented and so while your implementation may seem surprising - if those around you in business are good and are also totally confused by your moves you are likely running against a core competency. Systems design for predictability should be the most important principle of any operator. </ pontification>

But the madman theory guided the U.S. through the cold war?

> The opposite of unpredictable is strategic.

What does the noun 'strategic' mean here?

The author expands on what they mean by "strategic" in the following paragraphs.

I know, but I was wondering if that word is commonly used as a noun in English. I've only seen it used as an adj. Sorry should have been clear.

It is an adjective here. Both "unpredictable" and "strategic" are adjectives.

It is still an adjective in this context. English becomes more and more copula-dropping – i.e. leaving out the verb "to be" when it can be implied by context. If you add it back you can see how it's an adjective:

The opposite of (being) unpredictable is (to be) strategic.

Oh yea makes sense now. Thank you for your comment.

It's not a noun. I don't understand the logic you are implying.

Reply on news.ycombinator.com