rahaeli

We just flat out don't know which choice is the harm reduction choice. For other things like suicide, self-harm, eating disorder content, etc, there's enough evidence that we can create a harm-reduction-informed policy, but for this, there isn't :/

You have accurately identified one of the most significant topics of debate in the field, and like I said, we have almost no evidence either way. It's incredibly frustrating as a provider who's trying to build policy from an explicit harm reduction framework!

But psychiatrists who treat people who are seeking help to manage their desire to sexually offend against children but who are genuinely horrified at the thought of actually doing so often talk about how there is almost NO evidence on what treatments are effective :/

So it suffers from the same problem as any analysis of the legal issues: anyone who is highly motivated to do it also is way more likely to be starting from motivated reasoning, and you have to really really carefully evaluate the source to see if you can identify the motive.

Yeah, the quality of the research is abysmal, because there are obvious massive ethical issues in conducting it (and it's almost impossible to get IRB approval, but over and above that, the ethical issues themselves are a deterrent to any ethical researcher).

But just implementing a straightforward PhotoDNA check is technically very easy, and the access (and the API) is free. (The image classifiers that detect novel images and suck the least aren't, but.)

Now, if you're trying to detect novel images that aren't in PhotoDNA yet (because PhotoDNA is a reactive system that can only detect specific, known images), that's harder; there are a few services that have trained image classification models to detect them but they all suck in various ways.

--so they can screen out any material that is detected by PhotoDNA from their libraries that there's, understandably, a bit of a barrier to registration. But even that's not too high l, just establishing you're a legitimate UGC site and not a front.

It's even easier now if you use Cloudflare, which AO3 does; they implemented a system to do the checking for you. The biggest roadblock to implementation is getting approval to use the system in the first place; there are enough groups that keep trying to get access--

(There are a bunch of nuances I'm glossing over here, but that's the gist of it.) There are reasons not to implement PhotoDNA scanning and there are reasons to implement it; it's a complicated risk analysis. But the actual implementation of it is very very simple! They make it as easy as possible.

2258A does not require sites to proactively scan uploads; if scanning were mandated by law, it would turn the site into a government actor for 4th amendment analysis. The voluntary basis of it is the only reason, eg, PhotoDNA is not considered a "search" that requires a warrant.

18 U.S. Code § 2258A - Reporting requirements of providers

The full list of laws that trigger reporting and preservation requirements is in www.law.cornell.edu/uscode/text/... (specifically, (a)(2).) So image hosting doesn't really trigger *too* much additional compliance burden over and above what you're already required to do.

Also, even if the image isn't hosted on your servers, you are required to report if somebody posts something on your site with images embedded from elsewhere, and there's a bunch of other things you have to report that don't involve images at all!

I don't want to go into too much detail, but the only NCMEC report DW has ever had to make that was *definitely* CSAM was a user icon. (We've made one other, but that was borderline and we reported as a caution.)

Not a great environment to be in as a provider, I gotta say!

For a very long time, providers could be on a pretty solid basis with "if it's not photorealistic, we'll leave it unless a jury rules it is obscene" but now there's *almost* enough information to predict what they'll prosecute ... but only almost. And a lot of risk in getting it wrong.

And it's even more relevant because for a long time, prosecutions under §1466A were extremely rare, but they've been accelerating and there's been at least one that involved the site operator being charged (and convicted) for hosting it, sigh.

Seriously. There's just so many people who are so fucking damaged about this issue, sigh

FYI, I blocked my dude there because he's clearly just trolling

I am so, so thankful that I got out of working at scale before the advent of widely available handheld camera devices people carry with them 24/7 and if DW ever gets past a certain scale I'm going to go live in a cave in the desert

It is the only possible explanation I can think of

People reporting non-photorealistic artwork genuinely has material consequences on the efforts to actually help *actual child victims*. There are enough barriers in place already; the more people stop making it worse, the better.

Part of that overwhelmed is the number of reports NCMEC gets of non-photorealistic artwork! Providers are incentivized to over-report because that's the only way you get the liability shield, but it means NCMEC is fucking swamped!

NCMEC, the central clearinghouse for reporting, is so overwhelmed that they only have the ability to follow up on or refer out 0.02% of reports. The agencies they report out to drop even more. It's *incredibly* frustrating as a provider and I can only imagine how frustrated NCMEC analysts are.

I didn't get into the whole law enforcement thing, but, eg, the DOJ still hasn't finished fully implementing the cross-agency task force they were required to by law *in 2008* because of how many barriers there are in all this (and because of a lack of funding, sigh).

I just really think it's important in how we talk about this (and how people report it to websites!) to keep the focus where it belongs: is this evidence that an actual child is being/has been sexually abused? Because it's a highly important distinction!

"As of March 3, scientists have tallied 26 living chicks." ❤️

I will be SO THRILLED if the population hits 300 this year, seriously. It's such a wonderful conservation story! Thanks for the heads up!

An elderly calico cat rolled up into a ball with her face tucked under her paw

Boosted ❤️

Pet pic spam signal, ACTIVATE. @asivilezee.bsky.social is a sweetheart and having a rough time, she could use some cuteness.