Eight to Late

Sensemaking and Analytics for Organizations

Complex decision making as an infinite game

with 10 comments

A decision is the act of choosing between two or more options.

There are two kinds of decisions, computable and non-computable [1]. In the former, options are well-defined and finite in number, and there are unambiguous facts (data) available based on which options can be rated. In the latter, options are neither clear nor enumerable and facts, if available at all, are ambiguous.

Computable decisions are simple, non-computable decisions are complex. We’ll refer to the two decision types by these names in the remainder of this article.

An example of a simple decision is buying a product (TV, car or whatever) based on well-defined criteria (price, features etc.). An example of a complex decision is formulating a business strategy.

It should be clear that simple decisions involve smaller temporal and monetary stakes – i.e. the cost of getting things wrong is limited and the effects of a bad decision wear off in (a relatively short) time. Neither is true for complex decisions: the cost of a poor choice can be significant, and its negative effects tend to persist over time.

A key feature of complex decisions is that they (usually) affect multiple parties. That is, they are socially complex. This has implications regarding how such decisions should be approached. More on this later.

Conventional decision theory is based on the notion of maximizing benefit or utility. For simple decisions it is assumed that utility of each option can be computed; for complex decisions it is assumed they can be estimated, or at least ranked. The latter assumption is questionable because each party affected by a complex decision will have its own notion of utility, at least at the outset. Moreover, since neither options nor facts are unambiguous at the start, it makes little sense to attempt to estimate utility upfront.

The above being the case, it is clear that complex decisions cannot be made on the basis of maximizing utility alone.  Something else is needed.


James Carse’s classic book, Finite and Infinite Games, begins with the following lines:

There are at least two kinds of games. One could be called finite, the other infinite. A finite game is played for the purpose of winning, an infinite for the purpose of continuing the play.

A finite game ends when a player or team wins. However, “just as it is essential for a finite game to have a definitive ending, it must also have a precise beginning. Therefore, we can speak of finite games as having temporal boundaries.”

The parallel between simple decisions and finite games should be evident. Although less obvious, it is useful to think of a complex decision as an infinite game.

When making a complex decision – such as a business strategy – decision-makers will often focus on maximising potential benefits (aka utility). However, as often as not, the outcome of the decision will fall far short of the expected benefits and may, in some cases, even lead to ruin. This being so, it is perhaps more fruitful to focus on staying in the game (keep playing) rather than winning (maximising utility).

The aim of a complex decision should be to stay in the game rather than win.

How does one ensure that one stays in the game? Heinz von Foerster’s ethical imperative offers an answer”

Always act to increase your choices.

That is, one should decide in such a way that increases one’s options in the future thereby increasing chances of staying in the game. One can frame this discussion in terms of adaptability:  the greater the number of options the greater the ability to adapt to unexpected changes in the environment.

How can one “act to increase one’s choices”?

One way to do this is to leverage social complexity: get different parties to articulate their preferred options. Some of these options are likely to contradict each other. Nevertheless, there are ways to handle such a diversity of potentially contradictory views in an inclusive manner (for an example, see this paper; for more, check out this book). Such an approach also ensures that the problem and solution spaces are explored more exhaustively than if only a limited number of viewpoints are considered.


The point is this: there are always more options available than apparent. Indeed, the number of unexplored options at any stage is potentially infinite. The job of the infinite player (decision-maker) is to act so as surface them gradually, and thus stay in the game.


Traditionally, decision-making is seen as a logical undertaking based on facts or data. In contrast, when viewed as an infinite game, complex decision-making becomes a matter of ethics rather than logic.

Why ethics?

The answer lies in von Foerster’s dictum to increase one’s choices.  By doing so, one increases the chances that fewer stakeholders’ interests are overlooked in the decision-making process.

As Wittgenstein famously said, “It is clear ethics cannot be articulated.” All those tedious classes and books on business ethics miss the point entirely. Ethical matters are necessarily oblique:  the decision-maker who decides in a way that increases (future) choices, will be acting ethically without drawing attention to it, or even being consciously aware of it.


Any honest discussion of complex decision-making in organisations must address the issue of power.

Carse asserts that players (i.e. decision-makers in the context of this article) become powerful by acquiring titles (e.g. CEO, Manager etc.). However, such titles can only be acquired by winning a finite game– i.e. by being successful in competitions for roles. Power therefore relates to finite rather than infinite games.

As he notes in his book:

Power is a concept that belongs only in finite play. To speak meaningfully of a person’s power is to speak of what that person has already achieved, the titles they have already won.

Be that as it may, one cannot overlook the reality that those in powerful positions can (and often do) subvert the decision-making process by obstructing open and honest discussion of contentious issues. Sometimes they do so by their mere presence in the room.

How does a complex decision-maker deal with the issue of power?

Carse offers the following answer:

How do infinite players contend with power? Since the outcome of infinite play is endlessly open, there is no way of looking back to make an assessment of the power or weakness of earlier play. Infinite players look forward, not to a victory but toward ongoing play. A finite player plays to be powerful; the infinite player plays with strength. Power is concerned (and a consequence of) what has happened, strength with what has yet to happen. Power will be always restricted to a relatively small number of people. Anyone can be strong.

What strength means is context-dependent, but the following may help clarify its relationship to power:

Late last year I attended an end-of-year event at the university I teach at. There I bumped into a student I had mentored some time ago. We got talking about his workplace (a large government agency).

At one point he asked, “We really need to radically change the way we think about and work with data, but I’m not a manager and have no authority to initiate changes that need to be made.”

“Why don’t you demonstrate what you are capable of? Since you are familiar your data, it should be easy enough to frame and tackle a small yet meaningful data science problem.” I replied.

“What if my manager doesn’t like my taking the initiative?”

“It is easier to beg forgiveness than seek permission.”

“He might feel threatened and make life difficult for me.”

“If management doesn’t like you’re doing, it’s their loss. What’s the worst that could happen? You could lose your job. With what you are learning at university you should have no trouble moving on to another role. Indeed, by doing so, you will diversify your experience and increase your future options.”


To summarise:  when deciding on complex matters, act in a way that maximises possibility rather than utility. Such an approach is inherently ethical and enhances one’s chances of staying in the game.

Complex decision making is an infinite game.

[1] There are many other terms for this classification:  tame and wicked (Horst Rittel), programmed and non-programmed (Herbert Simon), complicated and complex (David Snowden). Paul Culmsee and I have, perhaps confusingly, used the terms uncertain and ambiguous to refer to these in our books.  There are minor contextual differences between how these different authors interpret these terms, but for the most part they are synonymous with computable/non-computable.


Written by K

January 21, 2020 at 4:09 am

10 Responses

Subscribe to comments with RSS.

  1. Very good. Thanks

    Liked by 1 person

    Tiba Delespierre

    January 21, 2020 at 9:49 am

  2. “Always act to increase your choices.” Ironically, benefits and utility, while not the decision-making goal, are outcomes resulting from successfully increasing one’s choices. If there were a ‘unified field theory equivalent’ for complex decision-making it is quite possible that acting to increase choices is an essential maxim. Thanks for articulating it so well. I believe I always approached complexity with increased possibilities being my aim but it wasn’t always an easy sell in the Fortune 100 company in which I worked. The few who understood it intuitively were rare. But, the bond created in the recognition of a shared mindset was a nice gift.

    Liked by 1 person

    Alexandre M.

    January 21, 2020 at 10:26 am

    • Thanks so much, Alexandre, I truly appreciate your taking the time to read this piece and for your generous comment. Totally agree that this approach is hard to sell in the corporate world where the focus is almost entirely on short term gain. Change, if it comes at all, will be slow and will happen through the efforts of the few who share the mindset you mention.

      Best wishes,




      January 21, 2020 at 10:44 am

  3. I found this interesting

    Two points come to mind

    1. The formulation of infinite games appears to be silent on the matter of the context or solution space being dynamic and altering in response to decisions we make. It might not have been the intention to assume that the unknowable options form a fixed backdrop, one we cannot see in its entirety but is largely static. Being silent on this issue that is what many readers will infer I think. The yearning for a ‘best’ decision and the assertion that the reason we cannot find it is the intractable number of options we face leaves out the idea that a system changes over time, not least in response to decisions we make, so that even if we had perfect knowledge of the current state of a system we would be unable to guarantee that a decision made now would be a good choice in the long term. If we acknowledge that the context will change when we make a decision or act on the system, we simply cannot guarantee that decisions made today will stand up in the long term. We might be lucky but constant vigilance and review of our strategies is the only way to cope with a complex emergent environment.

    2. As a result of the first point, while the concept of complexity described by Snowden and embodied in the Cynefin framework is non-computable, I don’t think the infinite games described here are entirely equivalent to complexity as described by Snowden. Complexity is a bigger concept than non-computable.

    The strategy of keeping options open makes sense anyway. I just feel there is more to be said and the complete acceptance of complexity, with its relationship to other ways of knowing a system (Cynefin), opens the way to choosing effective approaches when we seek to influence or manage a system.

    Liked by 1 person

    Steve Grey

    January 21, 2020 at 3:19 pm

    • Thanks for reading and for your thoughtful comments, Steve.

      The key feature of an infinite game is that the rule change in response to changes in the game. A couple of quotes from Carse may help clarify:

      The rules of a finite game may not change; the rules of an infinite game must change.


      Finite players play within boundaries; infinite players play with boundaries.

      Taken together, I read this as meaning two things for complex decisions – a) an acknowledgement that theunderstanding of the context of the decision problem is liable to change and b) the problem /context itself may change. In hindsight, I should have mentioned this explicitly in the article.

      To your second point: approaching a complex decision as an infinite game says nothing about how to play the game (i.e. make a provisional decision) or the nature of the problem – indeed the rules and context that define the problem are subject to change. One might use Cynefin or any other framework / facilitation method that is appropriate to the situation at hand. Indeed, I tend to use many. Framing a decision as an infinite game is more about the disposition (of the decision-maker(s)) than the techniques / frameworks employed to make a (provisional) decision, or even the nature of the problem itself.

      Liked by 1 person


      January 21, 2020 at 9:34 pm

  4. When I was a teenager, my father also emphasized that I should keep trying to increase my options, and that further education was a way of increasing my options. I think he told me how he came to see the importance of that principle, but I can’t remember the details now. (He didn’t learn it from von Foerster.)

    One part of what you wrote that I would question is: “All those tedious classes and books on business ethics miss the point entirely. Ethical matters are necessarily oblique: the decision-maker who decides in a way that increases (future) choices, will be acting ethically without drawing attention to it, or even being consciously aware of it.”

    Although I have never taken a class on business ethics, I have read a large sample of the literature, including dozens of articles in the Journal of Business Ethics, and not all of it is tedious (though I will grant that one could easily collect a tedious subset of the literature), and I don’t think it misses the point either. Acting to increase options is an important principle but not the only important principle. And without those other (evolving) principles, acting to increase options will not necessarily produce ethical action. In other words, the principle of always acting to increase options is only “inherently ethical” if certain other principles are also presupposed. I suspect that the reason why you think that acting to increase options is inherently ethical (and the literature on business ethics is irrelevant) is because you implicitly presume other principles that you don’t articulate here. I would have to say much more to adequately support that idea, but for now I just want to throw it out there as an objection.

    Liked by 1 person


    March 29, 2020 at 12:39 pm

    • Great to hear from you again Nathan, and thanks for picking up on my deliberately provocative statement! Indeed, I was hoping someone would :).

      I’m working from the Wittgensteinian position that ethics are are impossible to articulate and are therefore expressible only through action. Nevertheless, I completely agree that I’m likely presuming other principles that I have not articulated. I’d love to hear what you think they might be…and also some pointers to some of the non-tedious literature on business ethics.

      On a related note, I have come across really good pieces on value judgements and ethical choices in systems design, and have blogged some of them here. Some examples:




      There some good stuff on critical management studies as well. For example, this text by Mats Alvesson and Hugh Willmott:


      I’m yet to find anything of that nature dealing with business ethics, but I admit I haven’t looked too hard.

      Thanks again for your comment. Always a pleasure to hear from you.

      Hope you’re keeping well. Take care.



      March 29, 2020 at 1:56 pm

      • Kailash, thanks for the response, and thanks for the links to your other posts, which do articulate some of your other ethical presuppositions.

        You asked what some of the other important ethical principles might be. I think you already suggested one of the big ones in your post but you stated it negatively and seemed to imply that it necessarily follows from or is coupled to the “Always act to increase your choices” principle. You stated it as: “fewer stakeholders’ interests are overlooked in the decision-making process.” Stated positively it would be: “Always act to increase the number of stakeholders whose interests or needs are accommodated.”

        I suspect these two principles are independent because what increases options for some stakeholders may decrease options for other stakeholders, or even eliminate other stakeholders from the game, which doesn’t turn an infinite game into a finite game—it’s still an infinite game but not a maximally ethical one. In other words, the “your” in “Always act to increase your choices” can have a varying scope (that is, how big is your conception of “you”?), so at least a second principle (about the scope of the “your” in the first principle) is required to more fully define what is “ethical”, and even to constrain the first principle by decreasing the total field of options according to a criterion.

        What both principles have in common is “thinking bigger”: thinking about more options and more stakeholders. There are various conceptual frameworks that break down this idea of “thinking bigger” into levels of increasing scope; one that I have found to be entertaining is Bill Torbert’s “developmental action-logics”, as found in his book Action Inquiry: The Secret of Timely and Transforming Leadership. In that book Torbert explicitly connects his conception of action-logics with Carse’s book; Torbert wrote:

        “But most organizations, like most people, evolve through each of these early developmental transformations by a process of more or less traumatic trial and error that is never named, never explicitly recognized as a transformation of assumptions and action-logics, and never undertaken with the intention of eventually establishing a learning organization. In his book, Finite and Infinite Games, James Carse (1986) writes: ‘A finite game is played for the purpose of winning, an infinite game for the purpose of continuing the play. Finite players play within boundaries; infinite players play with boundaries.’ This imagery is apt in describing the evolving action-logics.”

        I have learned from many different general ethical orientations, but the general ethical orientation where I feel most at home is pragmatism. This orientation has been so far outside of the ethical-theory mainstream that most professional philosophers probably wouldn’t be able to tell you anything about it if you asked them. John Dewey (1859–1952) is arguably the central classical figure of this orientation (a good short secondary source is Jennifer Welchman’s 2010 chapter “Dewey’s Moral Philosophy” in The Cambridge Companion to Dewey, and a good longer secondary source is Gregory Pappas’s 2008 book John Dewey’s Ethics: Democracy as Experience). A pragmatist moral philosopher whom I recently discovered and have enjoyed reading is James D. Wallace, whose most recent book is Norms and Practices (2009). (He was also the father of the famous novelist David Foster Wallace.) This may have some affinity with Wittgenstein (some people have spoken of a “Wittgensteinian pragmatism”) but Deweyan pragmatism wouldn’t claim that (as you said) “ethics are are impossible to articulate and are therefore expressible only through action”, only that ethics are not fixed. Here’s a relevant excerpt from Welchman’s chapter that I just mentioned:

        “Consistent with his value pluralism, Dewey holds that there is no one ‘single commensurable principle’ that can be appealed to resolve problematic situations, individual or social. Nor should we expect any of our moral principles to ‘tell us’ what we should do. Since Dewey also holds that practical inquiry is continuous with natural scientific experimental inquiry, the principles it yields will be hypothetical, not categorical, and descriptive rather than normative in form. ‘The object of moral principles,’ Dewey writes, ‘is to supply standpoints and methods which will enable the individual [acting individually or collectively] to make for himself an analysis of the elements of good and evil, in the particular situation’ under review. That is, they are generalizations or generalized descriptions of relations between ends and means, practices and duties, dispositions and approbation, that we can use to determine what the obstacles to individual or collective endeavors are and what may be expected of any proposed solutions.

        “Every problematic situation is unique, but there are ‘generic features’ of human nature, situations, and outcomes, that lend themselves to generalization. These generalizations are both probabilistic and defeasible: they will fail to predict actual outcomes in a certain percentage of cases and fail to be applicable at all (i.e. ‘defeated’) when problematic situations deviate too far from the samples from which the generalizations were made. Likewise, every practice is unique, but there are generic features of practices we can capture in defeasible generalizations about what constitutes satisfactory performance of a practice, or satisfactory performance by a practice of its role within a set of social practices. And finally there are generic features of the admirable in human character traits that lend themselves to similar sorts of generalizations.

        “Commonsense morality is a vast repository of such principles to whose use our cultural training has habituated us. Being habituated to them, we can immediately and efficiently employ them at need, but are often so unconscious of them we give them little or no critical scrutiny. Since their role is descriptive and explanatory, they can be checked for their fruitfulness as analytical tools for assisting us in understanding problematic situations and predicting the outcomes of various kinds of responses. Principles failing these tests should be reassessed and revised accordingly. Because the roles, resources, and obstacles we meet with individually and collectively change over time, past assumptions about what should count as paradigmatic instances of any of these need periodic review….”

        I won’t overwhelm you with a long list of business ethics articles, but suffice it to say that there are plenty of them that are compatible with this orientation.

        Finally, here’s a quotation that I stumbled across as I was looking for material for this response. This is another quotation about Dewey, but this time from Joli Jensen, a professor of media studies who was writing about arts policy in her article “Expressive Logic: A New Premise in Arts Advocacy” (2003):

        “Dewey always argues for more—more variety, more engagement, more range, more liveliness. He presumes that we are all better off if we know and experience diversity in every aspect of our lives. For Dewey, multiplicity is good both in itself (he is a pluralist through and through) and for what it makes possible in modern life. Dewey, along with William James, values varieties of experiences. But Dewey, unlike James, makes the connection to democracy—in the end, he values varieties of experiences because they enhance our ability to participate in public life. Dewey understands modern democratic life as a field of possibilities in which citizens have responsibility and control. We must constantly choose our private and public destinies.”

        Liked by 1 person


        March 30, 2020 at 12:33 am

        • In my previous comment I said that pragmatism is “far outside of the ethical-theory mainstream” but on second thought I am not sure this is true anymore. For example, I just examined an ethics textbook, Craig E. Johnson’s Meeting the Ethical Challenges of Leadership: Casting Light or Shadow. In the “General Ethical Perspectives” chapter in the 2001 first edition, the book covers utilitarianism (“Do the most good for the greatest number of people”), Kant’s categorical imperative (“Do what’s right no matter what the cost”), communitarianism (“Shoulder your responsibilities/Seek the common good”), and altruism (“Love your neighbor”). The same chapter in the 2020 seventh edition drops communitarianism and adds justice as fairness (“Guaranteeing equal rights and opportunities behind the veil of ignorance”) and pragmatism (“Ethics as inquiry”). So over the course of the past two decades, pragmatism became mainstream enough to cover in this textbook.

          Liked by 1 person


          March 31, 2020 at 1:17 am

        • Thanks so much for this detailed follow up Nathan, in particular the pointers and quotes from Torbert and Wallace’s books. Both on my reading list now.



          April 2, 2020 at 7:45 am

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: