After 4.0 update, exploding gems only provides 50% mana. I’m totally fine with.
However, after examination, I discover that even the rounding formula is changed as well.
Pre 4.0, it’s simple rounding, rounded to the nearest integer, for instance, 0.7x1=1, 0.7x2=1.4=1, 0.7x3=2.1=2, 0.7x4=2.8=3, 0.7x5=3.5=4, and so on. (Blue line, x-axis being Gems of single color destroyed by explosion, y-axis being mana gain from such explosion)
Post 4.0, it’s banker’s rounding, rounded to the nearest even integer, for instance, 0.5x1=0 , 0.5x2=1, 0.5x3=1.5=2 , 0.5x4=2, 0.5x5=2.5=2 , and so on. (Red line)

If explode gems on larger scale, in fact mana gain is not so bad at all. It is 30% less mana gain on average, which is close enough to theoretical value of (0.7-0.5)/0.7=28.57%.
However, it’s absolutely abysmal on small scale. Weapon upgrade with explode one random gem now provides little to no mana gain. Even Fire Bomb could provide no mana gain. It’s 4 Red Gems exploded.

I would suggest make it round to nearest odd integer, then mana gain would be 0.5x1=1, 0.5x2=1, 0.5x3=1.5=1, 0.5x4=2, 0.5x5=2.5=3. This still provides 27.27%, close enough to 28.57%, less mana gain overall, but makes explode 1 gem still viable. (Green line)

PS: Already posted in other thread, but in Gameplay Chat category. I guess this needs more attention.

@saltypatra can devs confirm that the rounding changed? If so, why?

This seems to mean that single explodes now do basically nothing, which is an undisclosed and unnecessary structural nerf. If so, please can this be reversed ASAP…

Edit: to the OP, why use a contrived odd/even rounding at all? Simple rounding will be just fine, and far more understandable and predictable.

Because now it’s 50%, which always results in a 0.5 or 1. Simple rounding will bias towards higher value, and that would be too generous for Dev to implement.

Brought up the same thing regarding the rounding down of explorers.

Are you sure it didn’t always use Banker’s Rounding, though? The only way to even tell the difference would have been certain amounts of mana gathered while a troop had disease, or exactly 15 gems of a single color exploded (which would have resulted in 10 instead of 11) with the next time it would have made a difference between being 35 gems of that color (highly unlikely). Makes more sense they changed the values of a single variable without caring about the underlying mechanics they use for mana gathering and forgot to test low yield explodes than they changed. Especially after reading the official explanation that was given for why they were nerfing exploders.

If they didn’t change mana gathering math itself to fit this nerf and this is just a case of carelessness, the simplest fix would be to just change their exploderManaRatio variable to 0.51. The extra one mana per color 25% of the time is rarely going to be make-or-break for a given mass exploder, but it is everything for small scale exploders. This would result in low yield explodes being about where they were pre-patch until they hit at least 4 gems of a single color exploded, but they never really needed to be hit by the nerf in the first place, so win-win.

My understanding is that GoW uses Banker’s Rounding for everything and always has, it’s just that it wasn’t an issue much for 70% but has a much more frequent impact at 50%.

It’d be nice if a single gem didn’t give zero mana, but I suspect the devs took this into account when they decided on 50% being the reduction they wanted.

No. While explode gems still provides 70% mana gain, it’s simple rounding.
If it’s banker’s rounding in the first place, it would always be 0 mana gain on single gem destroyed by explosion, which isn’t the case at all.

For science, I checked this on delve chests. They use simple rounding, not Banker’s Rounding. From extensive testing, I’m fairly certain the number of drops you get is determined by chest level, and they round each drop, not total drops.

To double check, I had a level 4 chest with exactly a 2.5 modifier and ran away from the delve, producing:

Which is 1 legendary ingot, 1 mythic ingot, 20 glory, and 5 shards, each multiplied by the treasure modifier and rounded. In this case, 0.5 being rounded up to an odd number three times. I have another drop with a 3.5 modifier from earlier that yielded multiplies of 4 ingots, so clearly no bias to odd or even here.

But I’m still fairly convinced they didn’t even really consider low yield explosions here. After all, their explanation for the nerf was:

The last time magic increased was, what, January 2017? Unless we are counting the one point for Zuul’Goth owners. Exploders dominance couldn’t possibly be related to cascades prevalent on the Unity engine or even cascades related to oversaturation of storms. Nah, couldn’t be. Has to be all those points of magic we got in the past few months.

That isn’t how Banker’s Rounding works. Banker’s Rounding only differs from “standard” (“half up”) rounding for where to round decimal values of exactly 0.5. Where the “standard” form is “round half up”, Banker’s Rounding rounds half numbers toward the nearest even number. Which is why I specified the only numbers of exploded gems where Banker’s Rounding would have differed from “standard” rounding with a 0.7 explode modifier (without disease, since I’m not sure if disease rounds as another step or not) is 15 (11 with standard round up, 10 with bankers rounding) and 35 (25 with standard round up, 24 with bankers rounding). Every other multiple that lands on 0.5 that is possible on a single board without being matched away results in it already being rounded toward an even number (which is really only 5 and 25, 45 is not feasible).

Edit for bump. Didn’t mean to kill this topic. Single gem explode nerf is nonsensical with respect to both the stated reason for the exploder nerf and what actually caused (in many cases still causing) mass exploder dominance and should be addressed. I know thats bare minimum probably 3 months way, but if it isn’t being discussed now, it will probably never happen.