Hi There,
This is something that we need to test still (I encountered this in my own experiment - but only for when the task was run on mobile phones, for that reason I instructed all participants to use laptops and chrome where possible). If possible, can you track what trial number the crash usually occurs on? If it is predictable, perhaps the task could be split in two?
My suspicion is that the videos are clogging up the browser memory (which would also be why it ran ok other days, if the browser cache had time to clear!) - one possible solution you could test is if you can ask participants to clear their browser cache beforehand? (but I know that is not always ideal or possible).
I’de be interested to hear feedback on what type of device it occurs on, what browser, what trial number and how large each video is.
Hopefully we can debug this!
Becca