Linguistic Causes of Distracting Disagreement

· ·

There is disagreement which leads to constructive revision of definitions (see Plastic Definitions and Define/Destroy method), i.e., the improvement of definitions during innovation, and then there is disagreement which is distracting, useless, wastes time, and takes focus and attention away from improvements. Distracting disagreement comes from ambiguity, synonymy, and vagueness, what I call linguistic causes of disagreement that distracts.

Joel Sternfeld, Mclean, Virginia, December 1978 [Source]

Ambiguity, synonymy, and vagueness create risks which Plastic Definitions which we should be aware of, and even better, which we should proactively identify, estimate, and manage.

If two different words stand for the same idea, they are synonyms. A single word is ambiguous if it can stand for different ideas. It is clear that a noun can be ambiguous, or that two (or more) of them can be synonyms.

Same applies to verbs. One important verb in the logistics venture, which I mentioned in another text, was ”to match”. We wanted to build an online marketplace for freight transportation; if I grossly oversimplify, the marketplace is software which is used both by those who have freight and need to have it transported, and they are the demand side, and there is the supply side, those whose business is to transport freight. What a marketplace should do, is to match supply with demand, which is to say, make sure that as many businesses as possible from the demand side find someone on the supply side to move their freight.

When we started thinking about the market, we had to say how exactly we wanted demand to match to supply. “Matching” became a word we used hundreds of times daily, for months. It meant very different things to different people on the various teams. Very early on, when only six of us were involved in designing this market and its marketplace, I was responsible for proposing and analyzing different ways that ”matching” could work. At the very first meeting on this, I remember laying out a few dozen ways matching could be accomplished, each with its own pros and cons; and this was actually a small subset of what can be done. There is a field in academic economics research focused on so-called market design which deals, among other, exactly with this kind of problem.

”Economists have lately been called upon not only to analyze markets, but to design them. Market design involves a responsibility for detail, a need to deal with all of a market’s complications, not just its principal features. Designers therefore cannot work only with the simple conceptual models used for theoretical insights into the general working of markets. Instead, market design calls for an engineering approach.” [1] 

Ours was exactly an engineering approach, informed by economic theory (especially Alvin Roth’s work [1], including that with Marilda Sotomayor [2] and Axel Ockenfels [3]) and business practices in the logistics industry.

Is ”matching” an ambiguous verb? It isn’t in general: in daily informal, general-purpose usage, it is probably not. Another way to think of it, is that it is not ambiguous enough to cause trouble, so you don’t think much about all it could possibly mean. But when we had to design the marketplace, the meaning it had for the people involved was critical, since we were all wanting to build the marketplace to work exactly according to one single idea of matching. Keep in mind that there was no place where we could look for an established idea of matching, of how matching should work in our specific marketplace. That is, it’s not that we were failing to agree on some given definition of ”to match”, but we had to make a new definition for it, and agree.

If one of us thought it should work in one way, another in another way, and this disagreement remained opaque when each of us used the verb ”to match”, then any agreement we thought we had was creating risks as we continued designing and building that marketplace.

What about adjectives? In their case, issues can come from vagueness. Gradable adjectives are vague; big, small, tall, fast, easy, high, low, and so on, are all vague. Such adjectives imply an ill-defined scale, whose units have no standard, universal definition, and there is no generally agreed upon threshold which cuts that scale up: I might think this car is big, but you could say it isn’t, and we could both have good reasons to stand by what we thought. Is a car big if it is longer than some specified length? Is it heavy because it goes over some specified weight? There are no such specifications in general; have a look at an encyclopedia, Wikipedia included, if you disagree – that’s where it would normally show up. Does it?

One of our goals was to make matching ”transparent” for our users. We spent a long time trying to agree what that should mean. Should it mean that we show the budget available on the demand side to the supply side? Does it mean showing the name of the supplier to the customer? Notice the complexity behind just these two questions: they touch directly the business model, and more specifically how the operator of this marketplace – that was us – create value from managing it.

We were disagreeing with each other in the logistics venture, sometimes substantially, on nouns, verbs, and adjectives. As we build sentences over these, problems of the pieces we put together won’t go away. In those sentences, adverbs, pronouns, determiners, prepositions, and conjunctions wouldn’t – just by being added in a mix – remove ambiguity, synonymy, or vagueness.

There is nothing wrong per se with ambiguity, synonymy, and vagueness. They cause risks, and if you accept to live with those risks, then there’s no need to worry. However, here is one example where we failed to appreciate the risk and suffered eventually. So-called ”onboarding” of carriers is about how to bring (and again, I’m oversim- plifying) trucking companies to the supply side of the marketplace. The software was initially designed and delivered with a simple carrier onboarding process, the assumption being that the shorter it takes to the carrier, the better. It turned out that this was too optimistic; fraudulent carriers passed the process alongside trustworthy ones, and this caused issues for our customers and us. It took us months to regain trust with some of our pilot customers, and the launch of the software on the market also had to wait for repairs to relationships and code to be done. Besides the time of investors, management, engineering, and other teams, it was also a hit to morale, an aspect that does not lend itself easily to quantification.

Disagreement over the meaning of ”scalable” is another example. Investors kept insisting on having a scalable marketplace; we ended up designing and delivering software which could, based on simulations at least, scale to support all transactions in logistics in the North American market. But that was an unrealistic scale, one which no-one in this market could ever achieve (if only because of anti-trust regulation). Supporting that volume of transactions required a complex system, which was costly to change.Ambiguity, synonymy, and vagueness create risks which we should be aware of, and even better, which we should proactively identify, estimate, and manage. This is why with Plastic Definitions, it is necessary to be as precise, accurate, and clear as possible, even if it involves making premature design decisions – these help identify flaws, and iterate faster, thus powering the paradox at the core of the Define/Destroy method.

References

  1. Alvin E Roth. “The economist as engineer: Game theory, experimentation, and computation as tools for design economics”. In: Econometrica 70.4 (2002), pp. 1341–1378.
  2. Alvin E Roth and Marilda Sotomayor. “Two-sided matching”. In: Handbook of game theory with economic applications 1 (1992), pp. 485–541.
  3. Alvin E Roth and Axel Ockenfels. “Last-minute bidding and the rules for ending second-price auctions: Evidence from eBay and Amazon auctions on the Internet”. In: American economic review 92.4 (2002), pp. 1093–1103.