I use Matrix. I even kind of like Matrix and have high hopes for its future.

But like, good?

why would they?

People use it for things like planning terrorism, child enticement, and sharing CSAM. Discord probably wants to ban those people, and maybe aid in their prosecution.

Of course if you’re sharing something that’s actually private, Discord is a poor choice. So is email or DMs on Lemmy where the server admin could read the content.

Many people seem to be treating it differently now because the average person who catches it has milder symptoms than a couple years ago. The potential for long-term consequences seems to be left out of the conversation.

I won’t. That’s a terrible law and its use should never be celebrated, even against someone we don’t like.

How many people don’t know smoking is harmful to their health at the time they start? I think that fact has pervaded the public consciousness, yet people still start smoking.

Voting on Reddit was originally intended to be used as moderation rather than to express agreement or disagreement. That may be too subtle for a huge userbase.

Assuming USA - platform immunity (CDA 230) protects you from most liability for content originating elsewhere. There are specific requirements to maintain your immunity if you receive a DMCA takedown notice, or have actual knowledge that child pornography has been uploaded to your service.

Solved-ish.

I got webpack to run reliably by replacing its use of md4 with sha256 in these files:

$ grep -r md4 node_modules/webpack
node_modules/webpack/lib/ModuleFilenameHelpers.js:      const hash = createHash("md4");
node_modules/webpack/lib/optimize/ConcatenatedModule.js:                const hash = createHash("md4");
node_modules/webpack/lib/optimize/SplitChunksPlugin.js:         .createHash("md4")
node_modules/webpack/lib/NamedModulesPlugin.js: const hash = createHash("md4");
node_modules/webpack/lib/SourceMapDevToolPlugin.js:                                                             contentHash: createHash("md4")
node_modules/webpack/lib/WebpackOptionsDefaulter.js:            this.set("output.hashFunction", "md4");
node_modules/webpack/lib/HashedModuleIdsPlugin.js:                              hashFunction: "md4",

then in `config/initializers/content_security_policy.rb’, I replaced the line

.script_src :self, assets_host, "'wasm-unsafe-eval'"

with

p.script_src :self, assets_host, "'wasm-unsafe-eval' 'unsafe-eval'"

This seems like way more tinkering with the code and defaults than I should need to keep the server running so I’ll probably dig more later. I hope this post ends up being useful to anyone else having an issue.

refuse to download the reddit app, so I can’t see the original post

Rumor has it, Reddit has a website.

Which is why the reputation system can’t be based on something the user’s server says, but must be based on third parties the person checking the reputation trusts.

To give an example, @zaktakespictures@social.goodanser.com might claim to be a member in good standing at /c/photography@lemmy.world, having first posted 8 days ago, last posted today, posted 4 times in total.

You can check that manually by looking at the user page on lemmy.world and see that the posts were not removed by the community’s moderators, but you cannot check that the account is not banned as far as I know. What I have in mind would let your server query that sort of thing automatically and set up lists of communities you’ll trust to vouch for users.

There could be several options to deal with a user who doesn’t have reputation, such as not letting them post, holding their posts for moderation, or having a spam filter scrutinize their posts.

I don’t think it necessarily needs karma like Reddit, but I think a reputation system of some sort is going to be required for open federation to remain viable as federated systems grow. Just looking at account age and post history isn’t good enough if the bad actor owns a server and wants to put some effort into spamming or harassing people.