Here's a comparison of four versions of the game:
* the 8-bit ZX Spectrum version, using the Speccy emulator
* the 8-bit Commodore 64 version, using the CCS64 emulator
* the DOS version from here
* the GOG version.

https://i.imgur.com/Alj6wSo.png

For one, Ice Fear calculators don't seem to be the same as in the original, despite me repeating the same moves, and Luxor's position being the same. (And Morkin's, if it matters.)

More noticeably, the amount of killing done per day by both sides seems much lower in the GOG version. The starting conditions are the same (except the Ice Fear), and even though the enemy army might be bigger or smaller, it doesn't matter for your kill count because your troops attack before Doomdark's troops.

Chris Wild wrote about the Battle Algorithm on this page:
http://www.icemark.com/tower/ai.htm
But it's a bit confusing, and doesn't fully explain the numbers I'm seeing for ANY of these versions. (Note also the different ratios between warriors and riders).

So... how /exactly/ does the combat algorithm work, and why is it so different in the GOG version?

And what other game mechanics are different in this version?

BTW, is Chris Wild responsible for the GOG version as well? I saw his name in the manual. Does he read these forums? If you happen to see this post, first, thanks a lot for the dos version - it brought me lots of joy! Second of all, please tell us more about the combat algorithm! For the original game, too. I'm very curious about all the factors that go into it: do citadels and keeps really help, or is that just something the manual says? By how much? Does it work for both sides? Do Fey get better combat in forests? Riders vs warriors? etc etc.

Chris Wilde has been reading these forum, under the name icemarkuk.

Cool!

And I just did the same comparison in the Midnight Engine. The 1200 riders tend to kill about 500 while the 300 warriors kill a bit under 200, a 3:1 ratio of kills despite a 4:1 ratio of numbers. This even happens in the open, which I checked on the theory that a Keep favors warriors.

Also checked War of the Solstice on Frozen Empire. In that one, the warriors apparently didn't do any damage at all...

So I checked 6 versions of the game - half of them apparently by Chris Wild - and, apparently, every single one calculates battles differently. What's going on!

EDIT: found a third difference, and it's kinda important:
An unrecruited lord who loses a battle doesn't get displaced to an adjacent square. He just stands there until he dies, even if it takes a few days. Makes saving Korinel's sorry ass impossible if he gets tied up by a large force before you can recruit him.

Thanks for the reply. Haven't looked at that blog before; looks like I have some reading to do. But before I get to that, let me just quickly comment one thing: But you did notice that the warriors' damage output is also drastically different? Clearly shown in that screenshot in my first post.
In fact, I ran a series of tests this very morning. Lemme pull up the results... 300 fully invigorated warriors did, on average:
- 196 kills on Spectrum
- 179 kills in DOS
- 92.1 kills in GOG!

(I think the damage output of Doomdark's troops might be lowered even more severely in GOG, though I haven't tested for hard data so far. I feel weirdly safe walzing through enemy territory with a measly 1000 troops. Probably that Ice Fear issue, where I imagine relatively small differences become proportionately more noticeable when the Ice Fear value is low to begin with thanks to Luxor's presence. If it functions anything like a multiplier, then for example 15 vs 30 is a 2x difference, and it doesn't matter in the slightest if the total range is 0-255.)

On the other hand, Lords themselves are indistinguishable: Lord Blood does about 66 average kills in every version.

I've just started playing the GOG version myself, and I must say I was worried about the battle results. Lord Blood's 1200 Riders, who would usually killed 1275 of the Doomguard in the initial assault, only killed like 200-300.

Same with other lords; kill rates seem to be 1/4 of the original Spectrum version.
Doomdark's troops seem to deal the same damage as before.

This means the game will be harder, which is a plus, but it will also mean a much more drawn out affair, which is is a bigger minus.

I checked out the github version some time ago, 2.0.42 to be exact (sorry about late reporting, life has been distracting). I did observe a slight increase in riders' kills, but even with that, the overall number of kills by troops remains vastly lower than on Spectrum and in DOS, and still noticeably lower than on Commodore 64.

ETA: I was going to expand that screenshot I linked before (https://i.imgur.com/Alj6wSo.png) to include the newest version's results, but apparently I can't take screenshots in this game anymore. Neither printscreen nor even capturing video clips with Radeon ReLive does the job. Weird.

The GoG version was developed under an old framework that is no longer supported. I gave recently built new versions but as of yet have not been able to update GoG. In the mean time you can find newer versions on www.thelordsofmidnight.com/support - there have been a number of fixes for Doomdark's Revenge.