[mintplant]

> Fast forward to March of this year, to an entirely different story: the FBI arresting John Rivello for "assaulting" journalist Kurt Eichenwald with a tweet.

TechDirt phrases this as if to make light of the charge, but Rivello intentionally triggered Eichenwald's epilepsy by sending him a flashing GIF that read "YOU DESERVE A SEIZURE FOR YOUR POSTS". He followed this up with messages such as "I hope this sends him into a seizure", "spammed this at [Eichenwald] let's see if he dies", and "I know he has epilepsy". So what we're talking here extends beyond the typical case of Twitter harassment.

[Natsu]

I wonder how long it will be before someone develops a program that defends people from such content automatically?

I know there's been some research in the area, but I don't see much in the way of practical programs or screen filters:

https://phys.org/news/2009-07-software-tool-web-seizure-caus...

[throwanem]

If the trigger is only rapid alternation of frames with a high luminance and/or contrast delta, it seems like that could be detected fairly easily by methods not wholly dissimilar to those used in MPEG compression. But I'm not at all sure that is indeed the only trigger, and also not sure how that kind of analysis could be done close enough to realtime to make a usable display filter.

Might still be worth putting together as a library, though. I can see an easy win for sites like Twitter, which could apply it to user-submitted content before showing it to anyone.

[cwkoss]

Snowcrash IRL