On Discourse (open source forum software) we default new install to downloading hotlinked images so we can prevent this very specific problem.

But isn’t that copyright infringement?

Maybe? I'm not a lawyer and I would be surprised if they consulted one before implementing that feature. But unless the license for the file limits access it seems like they're just acting as an almost-proxy for these images, and not misrepresenting their ownership over them.

I'm not a lawyer, either, and have an honest question: how would a download of the image to cache it differ than a download to view it? They're performing the same action (downloading the bits). But, one stores it for later download by others, the second displays it (and maybe gets cached locally in a browser cache, making it look sort of like the first case...). Does intent matter? Is there an implied protection in that you can download for personal use, but cannot distribute? Does that break down on a multi-user computer? i.e. what if I download the image and my computer shows the cached image another user?

Well I believe it's a difference of redirection vs redistribution.

Say you're talking to someone who can teleport instantly across the globe. If you tell him where he can see the Mona Lisa, that's OK. If you grab the Mona Lisa and wave it in his face, that is very un-OK.

Before someone replies with the "but digital items can be copied, it's not stealing", that's the argument most piracy enabling forums and websites are able to get away with. If you host, EG, Nintendo ROMs on your website, you'll get shut down quickly. If you just save links (URL or Magnet), you might get a DMCA for a specific item maybe, but won't get shut down entirely, probably.

As always, there are a lots of exceptions and alternative interpretations though.

No. It’s called hot linking.

I always loved it when someone hotlinked to an image on someone else’s server and instead of taking the image down, they’d replace it with something funny (or more frequently, horrible).

The Web wouldn't be a reliable "web" if that behavior were widespread.

I loved it when the web was less reliable.

Something something about goats. ;)

Even with IPFS, you're pushing off infrastructure to someone who probably has some hidden incentive to provide that infrastructure.

Why not just charge the 100th visitor per minute or something? Someone will be motivated to pay the cost. And, if people don't want to pay? Well, even those with less motivation will find a way to the content somehow.