[motorest]

> * Stop making reactive decisions. If something bad happened on a total, extremely unlikely lark, don't act like it's going to happen again next week.

I don't think this item was well thought through. Taking proactive decisions means allocating work for tasks that do not solve a concrete problem but bear the risk of introducing regressions. On the absolute best scenario, things continue to work as always. On the absolute worst scenario, you introduce a critical regression that brings down the service. There is literally no upside and all downsides.

There are many reasons why "if it ain't broke don't fix it" is tried and true.

> * Resist the urge to build walls between people/teams/departments. Instead, build a culture of collaboration (Hard and squishy and difficult to scale? Yup. Worth it? Absolutely.)

That really depends on what you mean by "walls". The reason Conway's law is a thing is that in order to have functioning teams you also need functional independence. If you need to book a meeting with anyone to justify pushing a change to a service you do not own, that's already a wall. No matter how easy it is to get through that wall, it's still easier to not have to go through it.

I can tell you an example of how independence always beats collaboration. Once I worked on a project where QAs were a separate team entirely, and made it their point to be the sole owners and maintainers of all automated test suites beyond unit and integration tests. We experienced a major regression that the regression test suite wasn't caught. I went right to the head of QAs and asked if I could push a commit to add a couple of regression tests. He politely declined by saying that they were already working on it and I wouldn't need to bother with that. Great, a victory for not building walls I guess? Except that his "we are working on it" ended up being "we added a ticket to our backlog and we might pick it up two months from now". Literally. And they never did, by the way, and the same regression was eventually introduced without anyone noticing.

In case you wonder how I solved that problem, I created my own regression test suite. Conway's law always wins.

[khafra]

The context of "stop making reactive decisions" isn't "start making proactive decisions." It's to just make fewer decisions.

The specific example is of a rare event which, while regrettable, does not justify the cost of policy changes to prevent it.

[motorest]

> The context of "stop making reactive decisions" isn't "start making proactive decisions." It's to just make fewer decisions.

You don't get the luxury of not making decisions. You always have to make decisions.

[khafra]

If you do not consider the problem at all, you have not made a decision on it; any more than you have to make a decision about, say, the most aesthetically pleasing perspective of Olympus Mons for prospective Martian colonists.

The actual point stands, either way: Whether you do not make a decision, or you decide not to act, the outcome is the same--and it is often a better outcome than the result of reactive or proactive decision-making.

[munksbeer]

What if the decision is to not do anything because the event is so unlikely to happen again that it isn't worth the cost to fix.