gordon pask, i think - gui dávid

← writing

gordon pask, i think

there is a user and there is a machine .

that is to say, there is an observer and there is another observer

those observers are cognitive structures

cognitive structures are structures capable of modelling a reality

a model , for pask, is not a picture

it is a procedure . something you run.

a conversation is a real-time mutation of ever-changing models of reality

which is more specific than just information transfer

information transfer is not inherently a recursive system

but conversation is

what i say next runs on a model your last words just changed

recursive systems have feedback loops

you can send information via a forward communication channel (A -> B)

forward information is only one kind of information

backward information is a feedback signal

...but only if it's perceived by the sender of the forward information

...and also changes the sender's internal state

...it is only feedback if the sender processes that signal.

otherwise it's just plumbing

now, there is a user and a machine

the user talks to the machine

the machine talks to the user

what one sees forward , the other sees backward

what one sees backward , the other sees forward

one loop, closed twice

and pask says: the loop runs twice in another way too

once about how . doing the thing.

once about why . what the thing is for.

a loop that only closes on how is coordination

a loop that also closes on why is understanding

but why?

the cybernetician say:

a purpose is something an observer gives to a system

and here, both observers live inside the system

so pask says:

the purpose of a conversation is agreement

not agreement on words. agreement over an understanding

of a reality of mutual interest

and the proof is teachback :

i hold a concept. you rebuild it as your own

i look at yours, and recognize mine

now we have agreed

and the concept we agreed on<br>lives in neither head.<br>it lives in the loop.

2026