[0.17.17] PvP Scenario issues

[rain9441]

The following items are issues related to the PvP scenario. These items only apply when selecting New Scenario -> PvP.

#1: The starting tech does not apply team bonuses correctly. If you start with Red and Green tech unlocked, it will not have any inserter capacity bonuses nor mining productivity bonuses even though the techs are applied. It seems like the Force modifiers in the Balance Options screen is overriding those bonuses rather than adding them together. Researching additional technology after the map starts adds the tech bonuses appropriately.

To reproduce, start a new PvP scenario, check to make sure Starting Research Level is set to Space Science Pack and start the round. Open the Bonuses screen above the minimap and note that even though all tech is unlocked, the force has no bonuses to their items at all. May apply to worker robot speed/capacity, lab research, and train braking force as well.


#2: In the Production Score Calculator, the Cost, Gain, and Total values are not indicative of what they are. A fast inserter has a Cost of 92.05, a Gain of 100.6, and a Total of 8.55. The actual gain from building a fast inserter is 8.55 and the total is 100.6. Those labels seem to be backwards.

To reproduce, start a new PvP scenario, check to make sure Production Score is checked (right hand side) and start the round. Open the Production Score menu in the upper left and select a recipe in teh Recipe Calculator.


#3: In the Production Score condition, certain recipes are calculated incorrectly and produce significantly more points than they should based on the desired algorithm.

To reproduce this, start a new PvP scenario with Production Score enabled like the above. Open the recipe calculator and select Concrete. The recipe goes from 42 points to 94 points. Select Sulfuric Acid next; the recipe goes from 76 points to 121 points.

We (Syrcan) investigated a few things in the scenario and found that recipes that had more than 1 unit of output would have way bigger scores than they should. Turns out it was cause the energy addition was not taking into account the distribution. Randomdude has created a github project with the PvP Scenario in it and has created a pull request within it to show the issue and how it could be resolved.

The TL;DR is that any recipe with more than 1 entity in the output does not calculate the energy cost per output appropriately.
Possible solution #1 from Randomdude: https://github.com/randomdud3/factorio- ... ore/pull/2
Possible solution #2 from Randomdude: https://github.com/randomdud3/factorio- ... ore/pull/1

I am happy to work with Klonan directly to help these.

[randomdude]

The solution #1 breaks uranium as it takes into account the distribution of recipes, and there is a recipe that returns 99% of 238 and 1% of 235, making 235 have a really low value, while the reprocessing is the other way around, so as it takes minimum of all recipes, both end up with really low values, so maybe ignore that by the time being.


Also, raw-wood should be renamed to wood

[Klonan]

Thanks for the report,

I have fixed #1 for the next release.

#2 I don't think is a bug, the Sum-total for the recipe is a +10, the total being the gain - the cost

#3, Please make a post in ideas and suggestions. I am open to fixing the price calculation but I don't clearly see something that is working incorrectly from what you have written, so more information will be very helpful.

[Klonan]

I figures out the production score issue,

Before:


After:



This actually changes a lot of prices, but from what I can tell, its more correct.

As you said, it was related to the energy price addition, it was giving it as a flat value regardless of how much of the product was being produced.

So for instance,

Before:

1. 10 concrete = $100
2. 1 concrete = $10 + $10 of energy
3. 1 concrete = $20

Now:

1. 10 concrete = $100 + $10 of energy
2. 1 concrete = $11

[rain9441]

Thanks Klonan, that is the exact issue that we found with the algorithm. That covers issue! For what its worth, we did check out the prices after the change and did find a lot of differences. I believe oil refining ended up being a net negative across the board now. Oil and Uranium processing both have multiple recipes with varying outputs causing prices to be a little skewed. I wouldn't call those bugs though.

[Klonan]

Thanks Klonan, that is the exact issue that we found with the algorithm. That covers issue! For what its worth, we did check out the prices after the change and did find a lot of differences. I believe oil refining ended up being a net negative across the board now. Oil and Uranium processing both have multiple recipes with varying outputs causing prices to be a little skewed. I wouldn't call those bugs though.

I think the oil is now net negative because It picks the lowest price out of all available methods.

However, I was thinking it is maybe a nice fact that, it is some sunk cost into getting the refining up and running,
its not just a thoughtless choice to process the oil into the products without developing the later stages of production.