On Wed, Oct 12, 2016 at 2:54 AM, Aristotle Pagaltzis <pagalt...@gmx.de>
> The position of a czar is reactive and disempowered. They can only fire-
> fight individual issues as the project hurtles forward headlong.
> If you want real security or actual good design, it must be part of the
> culture: it must pervade every choice, every judgement call. Otherwise
> you get zilch, or maybe a veneer. Just the same is true of stability, or
> any other similar value.
Speaking personally again, I beg to differ. During the 2008 financial
crisis, when most banks were losing money, Goldman Sachs made money. Why?
One major reason was that the bank's risk management function was listened
to and heeded. All traders constantly *think about* risk, but they do so
in the context of tradeoffs in local decisions. Is the reward worth the
risk for this investment? Only the risk management function looked at the
aggregate risk of the bank as a whole and judged whether there was "too
much" risk. When risk management said "cut down those positions" or "hedge
that exposure", GS traders did what they said. Other banks treated risk
management like Cassandra in the myths. As the CEO of Citibank put it "as
long as the music is playing, you've got to dance."
The same could be true of a software project. Developers could be
assessing risk vs reward tradeoffs locally. Is *this* PR worth it? Will
*this* refactoring break too much stuff? There could well be a role for
stability oversight that looks at the aggregate whole and says "STOP". But
it only works if heeded by the rest – if the group trusts the oversight
enough to delegate the power to decide when it's too much.
Does it need to be a single person? Could it be a team responsibility like
a Toyota-style "Stop the Line" button? I think a community can decide how
it wants to mechanically run a stop process.
The important thing – if stability is truly important – is making sure that
people with the right perspectives on aggregate stability are known,
involved, and heeded.
David Golden <x...@xdg.me> Twitter/IRC/GitHub: @xdg
Searchable Archive: http://firstname.lastname@example.org