What Is the Foundation of the Seed for the Monk Conversion RNG?
- This topic has 8 replies, 1 voice, and was last updated 1 year, 9 months ago by
.
-
Posts
-
While watching a recent tournament, some comments were made about players’ luck with conversions.
This led me to wonder what the exact math is behind the conversion RNG and whether the seed used takes into account player names or IDs.
I don’t believe it uses local values like the host’s processor ID.
My initial guess is that it uses the Bcrypt.dll file that Windows stores its random number generator on, but I’m not sure if DE has their own unique calculation for this.
Some time ago I read a lengthy article about monk rng but I don’t recall where it was posted, here is what I remember (no guarantee that everything is 100% accurate).
Monk conversion works in cycles.
Each cycle has a certain chance of converting a unit.
Unfortunately I can’t remember if the chance is fixed or is increased with each cycle, I think it’s fixed though.
However there is minimum and maximum conversion times, meaning the first two or so cycles cannot convert.
For units with conversations resistence the minimum amount of cycles is higher.
For Mr Yo the minimum amount of cycles is cut in half.
At some point one cycle is guaranteed to convert when the maximum amount of cycles is reached.
Assuming that the game is using a halfway decent PRNG, the seed doesn’t matter.
There is no way to seed a (good) PRNG such that it consistently favours one or the other player.
Imo when people talk about players having (good or bad) “conversion luck”, one of two things is happening:
1.
The player has good monk micro, and had a conversion “charged up” against a target that ran away, before switching to a closer unit.
This is what happens in 95% of cases where people complain about “instant” conversions.
Alternatively, the player has bad monk micro and moves the monks, resetting the timer, or is just slow with starting the conversion.
2.It is 100% the bias of the caster, and any actual analysis of the randomness in all the conversions over all the games will show that every player actually has average luck.
From the Fandom Wiki
Conversion mechanics
From The Conquerors onwards, the Monk AI attempts to convert the target at regular intervals, called here conversion intervals (CI), roughly equal to 1 game-time second.
Once a Monk is at conversion range of his target (or touches it in case of rams, Armored Elephants Trebuchets, and buildings), his first CI starts.
His three first CIs have no effect on the target and only serve as warm-up.
Starting from the fourth CI (eight CI for the Scout Cavalry line and Eagle Scout line), the Monk has a 28% chance to convert a unit at the end of every CI.
If the Monk achieves “conversion success” in the warm-up phase, the target will be converted in the first actual CI irrespective of the “conversion chance” in that CI.
Even if the monk fails to convert by the maximum CI, the unit will be converted irrespective of the conversion chance in that CI.
Hence, the Monk is guaranteed to convert his target at the end of the 10th CI.
Attempts to convert buildings start at the 15th CI and are guaranteed to succeed at the end of the 25th CI, with all other mechanics being same.
In other words, Monks need between 4 and 10 CIs to convert a non-scout unit, between 8 and 10 CIs to convert a scout unit, and between 15 and 25 CIs to convert a building.
If multiple Monks are converting the same target, the conversion chances do not add up.
Example: If 4 or more Monks are converting the same target, conversion chance does not increase to 100%.
All conversion attempts are carried out independently, and if any of them succeed, the whole group succeeds.
Even if only one Monk succeeds, the whole group loses their faith, unless Theocracy was researched.
Switching targets
Switching targets does not reset the conversion intervals, even if done from unit to building or vice versa.
However, moving in any direction manually does reset it.
This makes it impossible to hit-and-run with Monks or Missionaries in the traditional sense.
If the Monk is chasing a unit automatically, his attempts of conversion are stored.
As such, when the target unit is back in the Monk’s conversion range, the process continues.
This also works with switching: if the target has fled, the Monk is chasing it automatically, and the player switches to some other enemy unit in the conversion range, the process works as if the Monk had been converting the new unit the whole time without taking a break.
Charging Monks
It is possible to charge Monks and get “instant conversions”, by abusing the above rules.
A group of Monks can attempt to convert a Scout Cavalry for 6 CIs (which will all fail since the unit cannot be converted before 8 CIs), and then switch to another unit, say Knight.
The Knight will be converted instantly if any of the 6 CIs resulted a conversion success.
A stronger method is by trying to convert buildings instead.
Since buildings cannot be converted before 15 CIs and units are always converted after 9 CIs, it is possible to get instant conversion every time on every unit if switching target between 10th and 14th CI.
The same happens after the 15th CI also, but then the possibility of conversion of the building also becomes real.
Effects of conversion resistance
All units start at resistance level 0.
In The Age of Kings, the Scout Cavalry had resistance level 3, and from The Conquerors to the Definitive Edition, the Scout Cavalry and Eagle Scout had resistance level 8, which was hardcoded.
Buildings start at resistance level 3.
The chance of conversion at any CI is given by 28/max(1,n)
%, where n is the conversion resistance level.The Teutons’ team bonus adds 1 minimum CI, 2 maximum CIs, and 2 resistance levels.
Faith and First Crusade add 2 minimum CIs, 4 maximum CIs, and 3 resistance levels.
The Bengalis’ bonus adds 1 minimum CI, and 1 maximum CI for elephant units.
Actually, units are not always guaranteed to be converted at the maximum CI (at least the very resistant ones).
The chance of conversion switches from 25 to 1,000, starting from the beginning of the maximum CI.
As such, for units with conversion resistance levels more than 10, it is possible that the unit is not converted even after the maximum CI.
Inquisition
Inquisition reduces minimum CI and maximum CI by 1 for all conversions and further reduces minimum CIs and maximum CIs by 5 for converting buildings (hence reduced by 6 in total for buildings).
Buildings are also slightly easier to convert by increasing the chances of conversion from 25 to 30 in the formula.
It affects both Monks and Missionaries.
Iirc T-West has a video on it
My guess is a custom random number generator, the same one used to generate the map.
The RNG is seeded at the start, identically for each player/spectator.
After the map is generated this RNG then also influences random events during game play: accuracy of arrow shots, monk conversions (I’m not sure if there is anything else).
This is integral to how the game functions online and it’s replays.
Each client must be able to reproduce the gameplay identically, given only starting conditions and player commands.
The seed has no effect on anything, it’s only a starting point.
YouTube has a few videos that talk decently in depth about this.
The community will vouch for Spirit of the Law, T-West, and Survivalist, (all of whom I respect), and there was another that built on what they had and highlights some really odd unique situations that can occur that break the rules
https://youtu.be/Xn_oYH4JBQk – Spirit of the Law, but not the video I wanted to find…
https://youtu.be/wPmiuAtsU08 – Survivalist
https://youtu.be/-qRUaOHpbwI – T-West
There is a lot of technique applied to monk micro that makes it look like you always get the good luck when it’s technique.
When you wonder why pros let monks wander off to chase light cav that is 2 screen away, it’s because they’re keeping that monk conversion primed.
You can also do that on buildings if you have redemption, and if you have atonement, it’s nice because you can do that on other monks, who are nice and slow and easy to keep up with.
If it’s a knight, they have limited sight and are a bit slower, so wait a bit before you start to convert them, and that allows you to build more conversion time.
Monks are a weird mix of you have to micro them a lot, but you can’t micro them nervously.
My pet theory is that Yo gets the good monk RNG because he is a more deliberate player who doesn’t frantically spam clicks every 1/4 second but makes good decisions.
I have a feeling that the fast players often are microing their monks too much and it’s actually working against them.
I feel like Hera had really good monk RNG in the latest tournament, as good as anybody, really.
I have a feeling he’s been working hard on that.
He’s really improved in the aspects of the game he was weak at a couple years ago.
Water maps used to be a free win against him.
At NAC4, he won every single match on Northern Isles.
It makes sense that he would have really studied Yo’s approach to monks and tried to get that good monk RNG himself.
- You must be logged in to reply to this topic.
