Wow there is an absolutely batshit concentration of ads on that page

From the linked paper[0], "In 2024, the entire GLM data record was reprocessed with new computational methods whose improved efficiency allowed difficult portions of the data record to be examined. This reprocessing inspired a reevaluation of the 2017 megaflash[...] after it identified a hitherto-overlooked discharge"

[0] https://journals.ametsoc.org/view/journals/bams/aop/BAMS-D-2...

By the animation it appears to be cloud to cloud.

Edit: I saw the animation in a different article. I'll come back if I can find it.

Space is 60 miles away, and this was over 500 miles. Maybe it hit the ground eventually, but the vast majority of the strike was definitely intra-cloud.

The simplest possible lightning bolt would be a straight line with a positive side and a negative side, no? If so, isn't that a single discharge?

Certainly what you're describing happens. You see the effect in the spark gaps of a Marx generator: one discharge triggers the next. I'm just saying that it's not necessarily the case that multiple differently charged branches mean spatially disconnected bolts. The path of least resistance is going to be through a nearby bolt of lightning, supposing one exists (plasma being a good conductor) so I'd expect they have a tendency to merge, so you get a network of sources and sinks. Probably as new nodes join the network what was once a source could become a sink--the channel being open for further equalization with new distant point.

Whether to count that as one thing or many, idk.

> it is actually multiple discharges

This might be the relevant sentence from the paper (but I'm not a weather scientist)

"Events were only used in the standard GLM clustering model to delineate flashes according to spatiotemporal proximity (events within 330 ms and 16.5 km of each other form the same flash)."