Why I am dubious that "better shared blocking" is even _an_ answer in the quest to make the fediverse #safEr:

1. Server Blocklists have their highest value in isolating Nazi instances and pro-CSAM instances. After you've blocked the known instances in this regard there's rapidly a question of what value is being provided ?

2. Meanwhile, once you get out of the "instances where no one on your service follows them" you start to impact relationships between uninvolved people.

1/

3. Blocks in general are _always_ reactive and shared blocklists _always_ have a delay in both additions and corrections

This means that your core shared blocklist is probably very useful! What people call your "tier0" list is quite powerful, especially when kept up to date

But as you get further and further away from tier 0 you get into increased collateral damage and lower value on the quality of the block. You also start to block instances that _by definition_ have done nothing to yours

2/

We use blocks—and blocklists—because we _lack better tooling_ and more fine grained controls.

Which on the one hand: sure, use the tools you got. But also be critical: is importing something that considers m.a's blocklist actually _providing meaningful value_? Is it actually preventing harassment? Could that reduction be obtained more easily via some other means or a smaller list?

On the other, if we are trying to create a systemically safer experience here, the value quickly declines.

3/

The value is higher—and the collateral damage is lower—from sharing things more fine grained than instance-instance level blocklists, or blocking for individuals.

_You_ choosing to use a shared blocklist, especially one that respects your follows, may have very high value even while your instance blocking another instance may not. There are still diminishing returns and a conversation to be had about false positives, but the calculus is _very_ different for a person.

4/

But what this means is that when we create tools to facilitate "better blocklist sharing" we need to acknowledge how that effort takes away from other potential solutions in this space.

Like as it is currently written, I'm pretty sure that the #FSEP proposal will not be mergeable into the #mastodon code base—they will balk at the increased centralization.

But there are a thousand small features could help, or could provide more value from the blocklists themselves, or lowering the costs.

5/

Basically: the more effort we put into tooling that _solves the wrong problem_ it takes away from our ability to contribute to tooling that is as likely (or more so) to move the needle, or that will make it more worthwhile to solve that original problem.

I'm just not convinced that adding additional features on top of instance-instance blocks helps here unless there is a dedicated focus on later steps and more robust tooling, which we seem to get distracted from while arguing blocklists.

6/

n.b. that when I say it's "solving the wrong problem" this isn't a dis on the people working in this space because I _do_ think there are interesting problems there that will eventually help…

It's just that, at least _in my view_, they aren't the problems that will make the fediverse as a whole safer. They may disagree and that's also fine, and I could very well be wrong, it just influences what we both choose to work on in this space, and who and how we recruit to various efforts.

7/7