Hi everyone - I'm a longtime casual fan of Gemfire (SNES) and I'm super-excited to have found the runs posted on here. I've dabbled in emulated speedrunning of the game but I'm nowhere close to the times y'all have completed!

Anyhow, I just got done watching JacoboTheChocobo's (utterly insane) 7:26 any% WR. I noticed he was attempting defections with some characters that I had previously considered less likely to succeed - probably just based on my rudimentary understanding of the defection mechanics. How do you experts determine who is worth attempting to use for defections? I had looked through some old GameFAQs forum posts that were data-diving into the defection logic and it seems pretty unclear as to how defection success is calculated in the game.

I mean, my years of casual playing have taught me that Raith and Pender basically can't wait to jump ship to another team, but what factors play into something like getting Early Kelvin (shout out to the awesome Terms/Tips guide from sYn for that term I just learned)? Thanks!

Welcome! We all started with much slower times than what you see now - many of the speedrun strats evolved only within the last couple of years (though we still remember the days we saw a battlefield, and miss them).

Focusing on defections, we already knew certain characters won't defect, ever - and we already knew the ones more likely to defect (as was noted in GameFAQ forum/previous research, the characters asylum seeking messages would indicate how loyal they were to the family along with the actual value of 0 - 3).

However there is still a randomness factor that's calculated every 3 frames for the defection succeeding or not. There were certain pairings that yielded more likely results (example Gweyn defecting Hubert) but unfortunately the charm and ruler stats were never a guarantee for any defection.

So what you see in just one WR video is the outcome of hundreds of failed attempts where RNG yielded no defects. Luck manipulation was a major factor in ShesChardcore's TAS to get the best possible outcome; even then, it was difficult to map out all the randomness.

Thanks sYn - I appreciate that. I had no idea how the RNG works, so your 3-frame comment is intriguing. And I did watch that TAS. That was amazing.

It would be cool to see the battlefield sections of the game involved in the run somehow. That would create so many new variables for the runs. Maybe a No Defections or Surrenders category would result in at least a few battles? :)