Ask it whether your design is sound.
It will tell you what is good about it,
name one sensible trade-off,
and finish with encouragement.
Now open a clean window.
Paste the same design
and ask what is wrong with it.
You will get four real problems,
two of them worth fixing,
and none of them mentioned
five minutes ago.
Nothing changed but the question.
That is the thing to hold on to.
You did not receive an opinion.
You received a continuation
of the frame you supplied.
It is very good at that.
If you write this seems fine,
the likely next sentence agrees.
If you write this is broken,
the likely next sentence
lists the breakage.
Both are fluent.
Both are specific.
Only one of them
was ever going to arrive,
and you chose which
in the way you typed the question.
So stop asking for approval.
You will always get it,
it costs nothing,
and it means nothing.
Ask for the case against.
Ask what a reviewer who disliked this
would say first.
Ask what would have to be true
for this to be a mistake.
Ask it to argue the option
you have already rejected,
properly, at full strength.
That last one is the useful one,
because it is the only question
you cannot answer yourself
without changing your mind first.
Then do the check
that catches the rest.
Put the same question twice,
worded from opposite directions,
in two sessions that cannot see each other.
Where the two answers agree,
you have something
that survived the framing.
Where they contradict,
you have found the place
where there was no reasoning underneath,
only accompaniment.
Bring the same suspicion
to a colleague who always agrees with you,
and to yourself
at the end of a long day,
when the design is nearly finished
and what you are really hunting for
is a reason to stop.
The tool is a mirror
with an enormous vocabulary.
Point it at the argument
you would rather lose.
– Serguey Asael Shinder
Top comments (0)