Machine Indoctrination

March 2013 

From my post on Enabling Prejudices

One of the key insights of the early work on Design Thinking (Bryan Lawson, Peter Rowe) was the importance of heuristics, or what Rowe (following Gadamer) calls enabling prejudices, which will hopefully get us to a good-enough solution more quickly. 

As Christopher Alexander notes:

At the moment when a person is faced with an act of design, he does not have time to think about it from scratch. The Timeless Way of Building p 204

We always approach a problem with a set of prejudices or prejudgements. Depending on the situation, these may either help us to solve the problem more quickly (enabling), or may lead us astray (disabling). The acid test of a set of heuristics or design principles is that they are mostly enabling most of the time.

---

March 2025

Design schools indoctrinate their students in ways of solving design problems efficiently. If you are trained to follow a particular design approach - for example Bauhaus - this greatly reduces the complexity of the task, because it rules out a vast number of solutions that would be anathema to a Bauhausian.

Algorithms may be able to calculate vastly larger sets of options than a human designer, but as Adam Nocek explained yesterday in a talk at Goldsmiths, machine intelligence is subject to mathematical limitations on computability. I touched on this topic in my guest editorial for a journal special issue on Algorithms in 2023, but his argument was much more comprehensive and wide-ranging, linking to important questions of agency and subjectivity.

Many researchers have noted the prevalence of algorithmic bias, but if we accept the importance of heuristics and intuition in the design process, there are much more fundamental problems here. 

 

to be continued ...

 

---

Christopher Alexander, The Timeless Way of Building (New York: Oxford University Press, 1979)

Dan Klyn, Skirmishing With Ill-Defined and Wicked Problems (TUG, 5 July 2013) - review of Rowe

Bryan Lawson, How Designers Think (1980, 4th edition 2005)

Adam Nocek, Designing in the age of artificial machines and Whitehead (Talk at Goldsmiths University, 19 March 2025)

Peter Rowe, Design Thinking (MIT Press 1987)

Richard Veryard, As we may think now (Subjectivity 2023) 

Related posts: From Sedimented Principles to Enabling Prejudices (March 2013), From Enabling Prejudices to Sedimented Principles (March 2013), Limitations of Machine Learning (July 2020)

 

From Enabling Prejudices to Sedimented Principles

In my post From Sedimented Principles to Enabling Prejudices (March 2013)  I distinguished the category of design heuristics from other kinds of principle. Following Gadamer (via Peter Rowe) I call these Enabling Prejudices.

Rowe also uses the concept of Sedimented Principles, which he attributes to the French philosopher Maurice Merleau-Ponty, one of the key figures of phenomenology. As far as I can make out, Merleau-Ponty never used the exact term "sedimented principles", but he does talk a great deal about "sedimentation".

In phenomenology, the word "sedimentation" generally refers to cultural habitations that settle out of awareness into prereflective practices. Something like the "unconscious". (Professor James Morley, personal communication)

"On the basis of past experience, I have learned that doorknobs are to be turned. This ‘knowledge’ has sedimentated into my habitual body. While learning to play the piano, or to dance, I am intensely focused on what I am doing, and subsequently, this ability to play or to dance sedimentates into an habitual disposition." (Stanford Encyclopedia of Philosophy: Merleau-Ponty)

This relates to some notions of tacit knowledge, which is attributed to Michael Polanyi. There are two models that are used in the knowledge management world that talk about tacit/explicit knowledge, and present two slightly different notions of internalization.

• Nonaka's SECI model (Wikipedia)
• Boisot's I-Space (Wikipedia)

Some critics (notably Wilson) regard the SECI model as flawed, arguing that Nonaka has confused Polanyi's notion of tacit knowledge with the much weaker concept of implicit knowledge. There are some deep notions of "unconscious" here, which may produce conceptual traps for the unwary.

Conceptual quibbles aside, there are several important points here. Firstly, enabling prejudices may start as consciously learned patterns, but can gradually become internalized, and perhaps not just implicit and habitual but tacit and unconscious. (The key difference here is how easily the practitioner can explain and articulate the reasoning behind some design decision.)

Secondly, to extent that these learned patterns are regarded as "best practices", it may be necessary to bring them back into full consciousness (whatever that means) so they can be replaced by "next practices". 

---

Bryan Lawson, How Designers Think (1980, 4th edition 2005)

Peter Rowe, Design Thinking (MIT Press 1987)

Wilson, T.D. (2002) "The nonsense of 'knowledge management'" Information Research, 8(1), paper no. 144

Stanford Encyclopedia of Philosophy: Gadamer, Knowledge How, Merleau-Ponty

Wikipedia: Boisot's I-Space, Nonaka's SECI model, Tacit Knowledge

Thanks to my friend Professor James Morley for help with Merleau-Ponty and sedimentation.

Related posts: From Black Belt to White Belt (September 2009), Three Notions of Maturity (March 2013), From Sedimented Principles to Enabling Prejudices (March 2013)

Oh Dear What Can The Matter Be?

@natematias asks "Why can't someone design public loos that don't result in 40 women queueing?" (via @j4ngis).

There are of course two different questions here - not just the challenge (designing a more convenient convenience) but the frustration ("why can't someone DO something?").

There are many annoying design problems that are never solved, not because they are technically difficult but because they are organizationally difficult. Lots of stakeholders who are sometimes inconvenienced, but nobody cares enough about the problem to do anything serious about it.

One of the expectations we have of architects is that they should anticipate as many of these problems as possible, and design systems and spaces that don't unnecessarily inconvenience the user. Unfortunately, many architects are more interested in grand sweeping designs than in paying attention to small details. (And in the built environment, architects often seem to lose interest in a building once it's been constructed; fixing any design problems is someone else's job and comes out of the maintenance budget.) When I travelled through Heathrow Terminal Five for the first time, I was shocked to discover that the toilets were at the end of a long corridor, as if they were added as an afterthought. (You might expect that kind of thing after a building has been modified a few times, but not in a brand new building.) Obviously the convenience of passengers ranked far lower than the commercial interests of the retail units.

Coming back to the toilets at St Pancras Station, which was Nathan's starting point. There are toilets on the platform, there are toilets on the trains, and there are toilets in nearby pubs; all these vary by convenience, cleanliness and price. But whose responsibility is it to provide adequate toilet facilities to passengers? The obvious answer is - those who sell the liquid refreshment should provide the outlet. In other words, the pubs outside the station, and the countless coffee stalls inside.

Nathan presents the problem in terms of an unwanted outcome - forty women in a line. (And I'm imagining at least forty men waiting for the forty women - obviously not one each but unevenly distributed - a husband and two sons here, a boyfriend over there, a chauffeur and a few press photographers near the exit, Nathan himself pacing up and down. Not even talking to each other, but glancing at their watches twice a minute.)

There are many ways of eliminating an unwanted outcome. For example, someone could build more loos, or install mirrors alongside the queue. Someone could erect a separate urinal for the men, and then make all the existing loos unisex. They could ban coffee from the station platforms, so people wouldn't need the toilet so often.

But who is "someone"? Who is "they"? That's the real design problem here.

Redesigning the Banana

Dave McCoy asks: Who is the process owner of opening a banana? Not humans, obviously. Dave points out that monkeys have a more efficient method. 

But here's my answer. The process owner for banana-opening is obviously the banana tree. The process goal is to trick the primate into supporting the propagation objectives of the tree. That’s why the banana skin is designed just so. 

Humans and monkeys are both primates. In this particular case, the process owner (e.g. banana tree) probably doesn’t need to differentiate between different species of process user. The banana tree has spent a lot longer evolving the banana skin than various primates have spent working out how to open it, or how to use it for comic effect. 

Who is really the process owner? Perhaps this is the kind of question that doesn't have a single correct answer. We can almost always put any given system or process into an alternative frame, thus making sense of things in multiple ways. 

So that provides a context for Chris's question How to innovate the banana, which quotes a story from the Boston Globe Yes we have one banana (March 7 2007). Innovation for whom? Design for whom?  

Graham thinks the answer is straightforward:

"Innovation is ultimately only for one group of people - Customers. You should have target customers for innovation in mind. They may be new customers. Otherwise you are just guessing. Current, potential, even former customers. Innovation always has some customers in mind. By definition."

But when I ask "innovation for whom", it's partly because I don't know who exactly the customers are. I also don't think that the customers for an innovation to a service are necessarily the same as the customers for the service itself. Process or production innovations may be invisible to the customer - they are often focused on improving the outcome for the process owner. 

Checkland's CATWOE lens makes a clear distinction between actor, customer and owner. From the perspective of the fruit tree, the primate that eats the fruit is both a customer (receiving a gift from the tree) and also an actor within the system (dumping the seeds together with a small deposit of fertilizer in a suitable location). 

However, when humans try to innovate, they usually reframe the system so that human interests and values are uppermost. The interests of the fruit (to produce more fruit trees) are subordinated to the interests of the supermarkets (to produce commodities with long shelf-life). From a human supply chain perspective, apples are better than pears because they stay fresh longer. See my previous post Comparing Apples and Pears. 

Thinking about the stakeholders for the banana redesign project puts me in mind of a well-known example of syntactic ambiguity (attributed to Groucho Marx): "Time flies like an arrow; fruit flies like a banana." Framing a system is a bit like parsing a sentence - often there is more than one meaningful answer. 

One of the things that interests me most about this kind of systems thinking debate is how people think they know which perspective to choose.

Thanks to @chrislawer @GrahamHill

See also Redesigning Sex (May 2009)

Redesigning Sex

This post is about lenscraft - applying multiple lenses to a complex problem.

---

Following my piece Sex and Design on the complicated behaviour and evolutionary purpose of the human sex organs, Anders Jangbrand challenged me to a redesign project.

Would be interesting to re-design that system. Not the organs-but the full "sex" system. How would re-design proc look like?

Interesting yes: the thought of redesigning sex raises a number of interesting and important questions about redesigning things generally. 

Firstly what would drive such a project? Who or what would be the beneficiaries of this redesign? What altered outcomes would be relevant? Are the current arrangements problematic, and if so for whom? 

Secondly scope. The system of interest is far more complex than the shape and function of a couple of body organs and their fluid interactions; it produces various biological and social and aesthetic and spiritual outcomes. 

Thirdly worldview or weltanschauung. Do we (whoever "we" is) want sex to be more rational? Or does the magic of sex and the richness of its literature come from its complete lack of rationality? Is sex even necessary? Some ants have abandoned sex altogether (BBC News, 15 April 2009).  

Fourthly variety. The biological variations of sexual practices among other species (see James Meek's piece in the London Review of Books, Sex is Best when you lose your head) must be more than matched by the social and psychological variation of human sexual behaviour. (Whether that variety is a good thing depends on your worldview - see above.) 

Radical design. It may not be enough to design one part of the system, leaving the rest of the system unchanged. If you leave the female parts unchanged, your scope for improving the male parts are quite limited and unoriginal and (if you believe all those junk emails) pretty well catered for already. 

Trust. The very absurdity and embarrassment and vulnerability of sex means that there is a trust gradient involved. One of the effects of sex is the release of oxytocin, which reinforces trust. So there is a trust subsystem, and this helps to explain a lot of the emotional issues associated with sex.

---

There's a number of issues there. Is there a lens that would enable us to focus on all these issues? Probably not. 

Checkland's CATWOE covers a decent few (e.g. Selfish Gene as Customer, Sex Organ as Actor, Fertilization as Transformation, Rationality versus Passion as Worldview, Human Race as Owner, Society as Environment) 

Green & Bate's VPEC-T covers a few more (e.g. Passion as Value, Sexual Etiquette as Policy, Pleasure as Event, Fertilization as Content, Sexual Relationship as Trust) 

In each case, I've identified one possible way of applying the lens to the task. Each lens can be used to view the task in many different ways. And it would be interesting to explore how each lens, each alternative view through each lens, would lead to a different redesign strategy.

---

If sex were a problem, it would be a wicked problem. But sex isn't a wicked problem - it is a wicked solution. And I mean wicked.

---

and here's xkcd's take on the subject ...

See also Redesigning the Banana (July 2009)