Towards intrinsically motivated systems of decision and action

I am going to talk about moti­va­tion, and I am going to talk about goal-set­ting. The two are relat­ed, of course. But when we aban­don instru­men­tal and deter­min­is­tic approach­es, it gets a lit­tle complicated. 

(If this post’s title sounds kind of scary, don’t take it too seri­ous­ly. I invent­ed it while role­play­ing an aca­d­e­m­ic, after writ­ing this in one pass.)


Because of my work at Hub­bub I read about moti­va­tion a lot. I remem­ber dur­ing the orig­i­nal gam­i­fi­ca­tion debates, a lot was said about intrin­sic moti­va­tion, and how arti­fi­cial exter­nal incen­tives actu­al­ly dimin­ish moti­va­tion. The evi­dence in sup­port of this keeps grow­ing, as described for instance in this recent piece in the New York Times. Here’s a quote:

Help­ing peo­ple focus on the mean­ing and impact of their work, rather than on, say, the finan­cial returns it will bring, may be the best way to improve not only the qual­i­ty of their work but also — coun­ter­in­tu­itive though it may seem — their finan­cial success.

Self-Deter­mi­na­tion The­o­ry (SDT) says that a basic human need is to feel autonomous. Extrin­sic incen­tives dimin­ish this sense of auton­o­my. In a work­place con­text, I can imag­ine that a dimin­ished sense of auton­o­my will lead to dimin­ished moti­va­tion to do good work.

I’ve been involved with quite a few work­place “gam­i­fi­ca­tion” projects (I con­tin­ue to dis­like the word but I’ll use it here for clar­i­ty’s sake). Our biggest chal­lenge was to get clients to decrease the amount of con­trol­ling feed­back already in place, in stead of adding even more under the guise of “fun”. This is the same thing that Kana­ga talks about when he talks about “soft gam­i­fi­ca­tion”.

The NYT arti­cle also talks about the dif­fer­ence between “inter­nal” ver­sus “instru­men­tal” motives. Inter­nal motives are inher­ent­ly relat­ed to the activ­i­ty at hand. Instru­men­tal ones are not. They lat­er dis­tin­guish internal/instrumental motives from internal/instrumental con­se­quences. If an activ­i­ty has instru­men­tal con­se­quences, it does not auto­mat­i­cal­ly fol­low that the per­son engaged in the activ­i­ty is moti­vat­ed by them. 

Going back to SDT, anoth­er need described is com­pe­tence, the sense of which is increased by offer­ing pos­i­tive feed­back. The study dis­cussed in the NYT arti­cle makes the impor­tant point that we should be look­ing for the inter­nal motives peo­ple have for engag­ing in a task, and help­ing them have a sense of inter­nal con­se­quences. It’s often eas­i­er to use instru­men­tal con­se­quences as the basis for our (dig­i­tal, gam­i­fied) pos­i­tive feed­back sys­tems, because they are ofte read­i­ly quan­tifi­able, and com­put­ers like stuff you can count. But this would actu­al­ly backfire.

In many ways, I am just rephras­ing stuff that has been said much bet­ter and more elab­o­rate­ly by Sebas­t­ian, and prob­a­bly also oth­ers. But it helps to hash these things out. It makes the con­cepts stick more.

Let’s shift.


In the land of pro­duc­tiv­i­ty, goal set­ting, par­tic­u­lar­ly of the SMART kind, is king. Indeed, in my own prac­tice at Hub­bub, one of the things we did when Alper became part­ner was to adopt Google’s OKR approach to goal set­ting to help us focus on what we want to achieve, and to pro­vide our­selves with feed­back on how we are doing. It’s not per­fect, but it works well enough and we con­tin­ue to use it.

But there’s a dan­ger to goal set­ting, or maybe a par­tic­u­lar kind of goal set­ting, which is nice­ly artic­u­lat­ed by Scott Adams, of all peo­ple, in a blog post titled “Goals vs. Sys­tems”. A quote:

My prob­lem with goals is that they are lim­it­ing. Grant­ed, if you focus on one par­tic­u­lar goal, your odds of achiev­ing it are bet­ter than if you have no goal. But you also miss out on oppor­tu­ni­ties that might have been far bet­ter than your goal. Sys­tems, how­ev­er, sim­ply move you from a game with low odds to a game with bet­ter odds. With a sys­tem you are less like­ly to miss one oppor­tu­ni­ty because you were too focused on anoth­er. With a sys­tem, you are always scan­ning for any opportunity.

Adams talks about set­ting your­self up to ben­e­fit from unex­pect­ed out­comes of the things you do. When we plan, and when we set goals, it can be tempt­ing to be very deter­min­is­tic in our approach. Adams sug­gests not focus­ing on goals but in stead cre­at­ing sys­tems that are gen­er­a­tive. When he says sys­tems, I sort of hear him say “habits”.

I think it’s more com­pli­cat­ed than aban­don­ing goals, though. Because the kind of sys­tems Adams sug­gests embrac­ing still serve goals, but like I said, in a less deter­min­is­tic way. He talks about increas­ing odds. And I think when he’s think­ing about those odds, he also has some poten­tial con­se­quences in mind.

This is basi­cal­ly a Talebian approach to goal-set­ting. It’s about mak­ing what Venkatesh Rao describes as “rich moves” (I can’t find the link to the par­tic­u­lar arti­cle I had in mind, alas).

The way I think about it for my own prac­tice of goal set­ting is to keep a loose cou­pling between the goals I want to achieve and the ways in which I expect to do so. I am basi­cal­ly look­ing for activ­i­ties (sys­tems) that get me clos­er to those goals, with­out decreas­ing the pos­si­bil­i­ty of oth­er good things hap­pen­ing too. It’s a game of trade-offs that starts from an accep­tance of the unpre­dictabil­i­ty of reality.

But what about motivation?


This is what I want to think about more. If we accept that moti­va­tion is best served by focus­ing on inter­nal con­se­quences. And if we believe that it is smarter (as in risk-savvy, not as in SMART) to focus on sys­tems in stead of goals, then how do we stay moti­vat­ed to dili­gent­ly walk through our sys­tems, in the absence of imme­di­ate pay­offs, or track­able progress towards a mea­sur­able goal?

This is per­son­al­ly rel­e­vant for me, as I am try­ing to get back on the blog­ging horse (sec­ond post of 2015, but it’s already week 4). It is also rel­e­vant because I want the OKRs we set at Hub­bub to be generative.

Maybe the moti­va­tion flow­charts Matt talked about way back when are help­ful here. And maybe Sebas­tian’s engage­ment loops are also use­ful. For now, the recipe I will be fol­low­ing for set­ting up “Adams sys­tems” that are intrin­si­cal­ly moti­vat­ed looks a lit­tle like this:

  • Under­stand the intrin­sic motives for engag­ing in the activ­i­ty at hand 
  • Deter­mine desired out­comes, both intrin­sic and instrumental
  • Brain­storm sys­tem-like activ­i­ties (habits) which increase the chances of these out­comes happening
  • Select the activ­i­ties which are most like­ly to have unex­pect­ed out­comes (or the least like­ly to have only expect­ed outcomes)
  • Invent ways of mak­ing appar­ent intrin­sic out­comes and reflect­ing on them
  • Loop back to your intrin­sic motives and adjust sys­tems accordingly

It’s a first stab, heav­i­ly inspired by Boy­d’s OODA-loop, which like I said before I am deeply into at the moment.

Let’s see how it works out, and let me know if it makes sense.

Published by

Kars Alfrink

Kars is a designer, researcher and educator focused on emerging technologies, social progress and the built environment.