6 replies to this topic

[Kanatta Jing]

Take a sample of 100 battles for each of the following eventualities in which Team Alpha (but not Team Beta I hate those guys) has zero, one or two (for example) Jenners.

Compare the wins versus the losses of those three situations.

Then we just mathemetize the information as follows.

One + Two - Zero = Value and we allow 2 full points per lance.

So if for instance Team A wins 48% of the time with Zero Jenners, wins 52% of the time when they have One Jenner and wins 60% of the time when they bring Two Jenners, then 0.52 + 0.6 - 0.48 = 0.64 or Jenners would be overpowered by 0.14.


Now then we just redo the math and reset the values every month or so.

Then we're adjusting the value based on the tactics in the community as well as rotating what the best mech is on a monthly basis prompting pilots to train a wide range of mechs, granting everyone a diverse playing experience and increasing players emotional and fiscal investment in the game.

Edited by Kanatta Jing, 03 August 2012 - 04:42 PM.

[CCC Dober]

What about using a fully automated combat simulator/routine/script instead?
It can't get more objective than that.

[Kanatta Jing]

CCC Dober, on 04 August 2012 - 05:19 AM, said:

What about using a fully automated combat simulator/routine/script instead?
It can't get more objective than that.

But reality isn't objective, players treating different Mechs differently changes their performance.

[CCC Dober]

If you want a correct assessment of performance, you need to be objective.
AI routines/constructs/players are just that. They don't have emotions, they don't feel, they simply fight to the death and can be matched in terms of skill so that it does not affect the outcome. Unless you have access to lobotomized humans that know how to play MWO and cage them in a testlab to log their results, I don't see a better/humane solution, honestly. Not to give you any stupid ideas, I hope the example is ridiculous enough to highlight the advantages of AI in this particular case.

[Exilyth]

Kanatta Jing, on 03 August 2012 - 04:39 PM, said:

Now then we just redo the math and reset the values every month or so.

So, you just have to play bad in your favorite mech for a month to get a month where that mech is undervalued?

[OriontheHunter]

Exilyth, on 04 August 2012 - 02:50 PM, said:

So, you just have to play bad in your favorite mech for a month to get a month where that mech is undervalued?

That or hope that a bunch of newbie's take care of that for you.

@OP CCC Dober is right that AI sims would be a better way to do this. It's the only way to accurately rate the machine, not the pilot.

[Kanatta Jing]

OriontheHunter, on 04 August 2012 - 07:13 PM, said:

That or hope that a bunch of newbie's take care of that for you.

@OP CCC Dober is right that AI sims would be a better way to do this. It's the only way to accurately rate the machine, not the pilot.

If the mech is over powered, but is also a confused mess that 99% of pilots can't use properly. Is it really all that valuable?

If a mech has an overly fearsome reputation and is constantly and near instantly killed does it really have an impact on the field?