Move 37

Design­ers make choic­es. They should be able to pro­vide ratio­nales for those choic­es. (Although some­times they can’t.) Being able to explain the think­ing that went into a design move to your­self, your team­mates and clients is part of being a pro­fes­sion­al.

Move 37. This was the move Alpha­Go made which took every­one by sur­prise because it appeared so wrong at first.

The inter­est­ing thing is that in hind­sight it appeared Alpha­Go had good rea­sons for this move. Based on a cal­cu­la­tion of odds, basi­cal­ly.

If asked at the time, would Alpha­Go have been able to pro­vide this ratio­nale?

It’s a thing that pops up in a lot of the read­ing I am doing around AI. This idea of trans­paren­cy. In some fields you don’t just want an AI to pro­vide you with a deci­sion, but also with the argu­ments sup­port­ing that deci­sion. Obvi­ous exam­ples would include a sys­tem that helps diag­nose dis­ease. You want it to pro­vide more than just the diag­no­sis. Because if it turns out to be wrong, you want to be able to say why at the time you thought it was right. This is a social, cul­tur­al and also legal require­ment.

It’s inter­est­ing.

Although lives don’t depend on it, the same might apply to intel­li­gent design tools. If I am work­ing with a sys­tem and it is offer­ing me design direc­tions or solu­tions, I want to know why it is sug­gest­ing these things as well. Because my rea­son for pick­ing one over the oth­er depends not just on the sur­face lev­el prop­er­ties of the design but also the under­ly­ing rea­sons. It might be impor­tant because I need to be able to tell stake­hold­ers about it.

An added side effect of this is that a design­er work­ing with such a sys­tem is be exposed to machine rea­son­ing about design choic­es. This could inform their own future think­ing too.

Trans­par­ent AI might help peo­ple improve them­selves. A black box can’t teach you much about the craft it’s per­form­ing. Look­ing at out­comes can be inspi­ra­tional or help­ful, but the process­es that lead up to them can be equal­ly infor­ma­tive. If not more so.

Imag­ine work­ing with an intel­li­gent design tool and get­ting the equiv­a­lent of an Alpha­Go move 37 moment. Huge­ly inspi­ra­tional. Game chang­er.

This idea gets me much more excit­ed than automat­ing design tasks does.

Orientation is the Schwerpunkt”

Putting this here so I can point to it. Such an impor­tant con­cept that has real­ly changed the way I approach deci­sion mak­ing. I used to oper­ate in some­thing like an observe-decide-act man­ner. But under­stand­ing that you can ori­ent to change your options is cru­cial for the abil­i­ty to win.

Ori­en­ta­tion is the Schw­er­punkt. It shapes the way we inter­act with the envi­ron­ment — hence ori­en­ta­tion shapes the way we observe, the way we decide, the way we act.

Ori­en­ta­tion shapes the char­ac­ter of present obser­va­tion-ori­en­ta­tion-deci­sion-action loops — while these present loops shape the char­ac­ter of future ori­en­ta­tion.

—John Boyd, Organ­ic Design for Com­mand and Con­trol

Blog All Kindle-clipped Locations: Destruction and Creation

I fin­ished my pre­vi­ous post on why design­ers should be inter­est­ed in John Boyd with the rec­om­men­da­tion to read his essay “Destruc­tion and Cre­ation”. I thought I’d share the bits I high­light­ed in my copy. It is part of Osinga’s Sci­ence, Strat­e­gy and War, to which the loca­tions below refer.

Loca­tion 3176 – Boyd intro­duces a very sim­ple but fun­da­men­tal rea­son for why we should care about deci­sion mak­ing:

… a basic aim or goal, as indi­vid­u­als, is to improve our capac­i­ty for inde­pen­dent action

Loca­tion 3183 – the same applies to design and design­ers. We do not want to be con­trolled by our cir­cum­stances. Boyd was talk­ing to a mil­i­tary audi­ence, but the descrip­tion below is true of any social sit­u­a­tion, includ­ing the design prac­tice:

In a real world of lim­it­ed resources and skills, indi­vid­u­als and groups form, dis­solve and reform their coop­er­a­tive or com­pet­i­tive pos­tures in a con­tin­u­ous strug­gle to remove or over­come phys­i­cal and social envi­ron­men­tal obsta­cles.

Loca­tion 3190

Against such a back­ground, actions and deci­sions become crit­i­cal­ly impor­tant.

Loca­tion 3192

To make these time­ly deci­sions implies that we must be able to form men­tal con­cepts of observed real­i­ty, as we per­ceive it, and be able to change these con­cepts as real­i­ty itself appears to change.

Loca­tion 3195 – design­ers are asked to do noth­ing but the above. The suc­ces of our designs hinges on our under­stand­ing of real­i­ty and our skill at inter­ven­ing in it. So the ques­tion below is of vital impor­tance to us:

How do we gen­er­ate or cre­ate the men­tal con­cepts to sup­port this deci­sion-mak­ing activ­i­ty?

Loca­tion 3196 – in the next sec­tion of the essay Boyd starts to pro­vide answers:

There are two ways in which we can devel­op and manip­u­late men­tal con­cepts to rep­re­sent observed real­i­ty: We can start from a com­pre­hen­sive whole and break it down to its par­tic­u­lars or we can start with the par­tic­u­lars and build towards a com­pre­hen­sive whole.

Loca­tion 3207

… gen­er­al-to-spe­cif­ic is relat­ed to deduc­tion, analy­sis, and dif­fer­en­ti­a­tion, while, spe­cif­ic-to-gen­er­al is relat­ed to induc­tion, syn­the­sis, and inte­gra­tion.

Loca­tion 3216

… such an unstruc­tur­ing or destruc­tion of many domains – to break the cor­re­spon­dence of each with its respec­tive con­stituents – is relat­ed to deduc­tion, analy­sis, and dif­fer­en­ti­a­tion. We call this kind of unstruc­tur­ing a destruc­tive deduc­tion.

Loca­tion 3225

… cre­ativ­i­ty is relat­ed to induc­tion, syn­the­sis, and inte­gra­tion since we pro­ceed­ed from unstruc­tured bits and pieces to a new gen­er­al pat­tern or con­cept. We call such action a cre­ative or con­struc­tive induc­tion.

Loca­tion 3227 – here Boyd starts to con­nect the two ways of cre­at­ing con­cepts. I have always found it grat­i­fy­ing to immerse myself in a design’s domain and to start teas­ing apart its con­stituent ele­ments, before mov­ing on to acts of cre­ation:

It is impor­tant to note that the cru­cial or key step that per­mits this cre­ative induc­tion is the sep­a­ra­tion of the par­tic­u­lars from their pre­vi­ous domains by the destruc­tive deduc­tion.

Loca­tion 3230

… the unstruc­tur­ing and restruc­tur­ing just shown reveals a way of chang­ing our per­cep­tion of real­i­ty.

Loca­tion 3237 – so far so fair­ly straight-for­ward. But Boyd gets increas­ing­ly more sophis­ti­cat­ed about this cycle of destruc­tion and cre­ation. For exam­ple, he sug­gests we should check for inter­nal con­sis­ten­cy of a new con­cept by trac­ing back its ele­ments to the orig­i­nal sources:

… we check for reversibil­i­ty as well as check to see which ideas and inter­ac­tions match-up with our obser­va­tions of real­i­ty.

Loca­tion 3240 – so this is not a two-step lin­ear act, but a cycli­cal one, where we keep tun­ing parts and wholes of a con­cept (or design) and test them against real­i­ty:

Over and over again this cycle of Destruc­tion and Cre­ation is repeat­ed until we demon­strate inter­nal con­sis­ten­cy and match-up with real­i­ty.

Loca­tion 3249 – in the next sec­tion, Boyd prob­lema­tis­es the process he has pro­posed by show­ing that once we have formed a con­cept, its matchup to real­i­ty imme­di­ate­ly starts to dete­ri­o­rate:

… at some point, ambi­gu­i­ties, uncer­tain­ties, anom­alies, or appar­ent incon­sis­ten­cies may emerge to sti­fle a more gen­er­al and pre­cise match-up of con­cept with observed real­i­ty.

Loca­tion 3257 – the point below is one I can’t help but iter­ate often enough to clients and cowork­ers. We must work under the assump­tion of mis­match­es occur­ring soon­er or lat­er. It is an essen­tial state of mind:

… we should antic­i­pate a mis­match between phe­nom­e­na obser­va­tion and con­cept descrip­tion of that obser­va­tion.

Loca­tion 3266 – he brings in Gödel, Heisen­berg and the sec­ond law of ther­mo­dy­nam­ics to explain why this is so:

Gödel’s Proof indi­rect­ly shows that in order to deter­mine the con­sis­ten­cy of any new sys­tem we must con­struct or uncov­er anoth­er sys­tem beyond it.

Loca­tion 3274

Back and forth, over and over again, we use obser­va­tions to sharp­en a con­cept and a con­cept to sharp­en obser­va­tions. Under these cir­cum­stances, a con­cept must be incom­plete since we depend upon an ever-chang­ing array of obser­va­tions to shape or for­mu­late it. Like­wise, our obser­va­tions of real­i­ty must be incom­plete since we depend upon a chang­ing con­cept to shape or for­mu­late the nature of new inquiries and obser­va­tions.

Loca­tion 3301 – so Gödel shows we need to con­tin­u­ous­ly cre­ate new con­cepts to main­tain the use­ful­ness of pri­or ones due to the rela­tion­ship between observed real­i­ty and men­tal con­cepts. Good news for design­ers! Our work is nev­er done. It is also an inter­est­ing way to think about cul­ture evolv­ing by the build­ing of increas­ing­ly com­plex net­works of pri­or con­cepts into new ones. Next, Boyd brings in Heisen­berg to explain why there is uncer­tain­ty involved when mak­ing obser­va­tions of real­i­ty:

… the mag­ni­tude of the uncer­tain­ty val­ues rep­re­sent the degree of intru­sion by the observ­er upon the observed.

Loca­tion 3304

… uncer­tain­ty val­ues not only rep­re­sent the degree of intru­sion by the observ­er upon the observed but also the degree of con­fu­sion and dis­or­der per­ceived by that observ­er.

Loca­tion 3308 – Heisen­berg shows that the more we become intwined with observed real­i­ty the more uncer­tain­ty increas­es. This is of note because as we design new things and we intro­duce them into the envi­ron­ment, unex­pect­ed things start to hap­pen. But also, we as design­ers our­selves are part of the envi­ron­ment. The more we are part of the same con­text we are design­ing for, the less able we will be to see things as they tru­ly are. Final­ly, for the third move by which Boyd prob­lema­tis­es the cre­ation of new con­cepts, we arrive at the sec­ond law of ther­mo­dy­nam­ics:

High entropy implies a low poten­tial for doing work, a low capac­i­ty for tak­ing action or a high degree of con­fu­sion and dis­or­der. Low entropy implies just the oppo­site.

Loca­tion 3312 – closed sys­tems are those that don’t com­mu­ni­cate with their envi­ron­ment. A suc­cess­ful design prac­tice should be an open sys­tem, lest it suc­cumb to entropy:

From this law it fol­lows that entropy must increase in any closed sys­tem

… when­ev­er we attempt to do work or take action inside such a sys­tem – a con­cept and its match-up with real­i­ty – we should antic­i­pate an increase in entropy hence an increase in con­fu­sion and dis­or­der.

Loca­tion 3317 – it’s impor­tant to note that Boyd’s ideas are equal­ly applic­a­ble to design plans, design prac­tices, design out­comes, any sys­tem involved in design, real­ly. Con­fused? Not to wor­ry, Boyd boils it down in the next and final sec­tion:

Accord­ing to Gödel we can­not – in gen­er­al – deter­mine the con­sis­ten­cy, hence the char­ac­ter or nature, of an abstract sys­tem with­in itself. Accord­ing to Heisen­berg and the Sec­ond Law of Ther­mo­dy­nam­ics any attempt to do so in the real world will expose uncer­tain­ty and gen­er­ate dis­or­der.

Loca­tion 3320 – the bit below is a pret­ty good sum­ma­ry of why “big design up front” does not work:

any inward-ori­ent­ed and con­tin­ued effort to improve the match-up of con­cept with observed real­i­ty will only increase the degree of mis­match.

Loca­tion 3329 – when­ev­er we encounter chaos the instinct is to stick to our guns, but it is prob­a­bly wis­er to take a step back and recon­sid­er our assump­tions:

we find that the uncer­tain­ty and dis­or­der gen­er­at­ed by an inward-ori­ent­ed sys­tem talk­ing to itself can be off­set by going out­side and cre­at­ing a new sys­tem.

Loca­tion 3330 – cre­ativ­i­ty or explo­rative design under pres­sure can seem like a waste of time but once we have gone through the exer­cise in hind sight we always find it more use­ful than thought before:

Sim­ply stat­ed, uncer­tain­ty and relat­ed dis­or­der can be dimin­ished by the direct arti­fice of cre­at­ing a high­er and broad­er more gen­er­al con­cept to rep­re­sent real­i­ty.

Loca­tion 3340

I believe we have uncov­ered a Dialec­tic Engine that per­mits the con­struc­tion of deci­sion mod­els need­ed by indi­vid­u­als and soci­eties for deter­min­ing and mon­i­tor­ing actions in an effort to improve their capac­i­ty for inde­pen­dent action.

Loca­tion 3341

the goal seek­ing effort itself appears to be the oth­er side of a con­trol mech­a­nism that seems also to dri­ve and reg­u­late the alter­nat­ing cycle of destruc­tion and cre­ation toward high­er and broad­er lev­els of elab­o­ra­tion.

Loca­tion 3347 – chaos is a fact of life, and as such we should wel­come it because it is as much a source of vital­i­ty as it is a threat:

Para­dox­i­cal­ly, then, an entropy increase per­mits both the destruc­tion or unstruc­tur­ing of a closed sys­tem and the cre­ation of a new sys­tem to nul­li­fy the march toward ran­dom­ness and death.

Loca­tion 3350 – one of Boyd’s final lines is a fine descrip­tion of what I think design should aspire to:

The result is a chang­ing and expand­ing uni­verse of men­tal con­cepts matched to a chang­ing and expand­ing uni­verse of observed real­i­ty.