plasmoid wrote:Mike has certainly suggested that (TV+) as a matching criteria and inducement generator for MM Leagues. As a side note he suggests that it Works the best in a no-perm-injuries setting. Which may well be a deal breaker for a lot people.
Either way, TV+ (given time) creates fair match-ups in an MM setting.
For MM it's better in a no permanent injury setting - that's because the "rez" setup deals with the massive and uninvested playerbase issue in which people are constantly attempting to game the system or play their own meta-games (eg, griefing). In other environments that's not much of an issue and thus, doesn't require that sort of solution.
TVPlus is better than TV in all environments for everything TV is used for. Full stop. It's an easy claim to make since TV isn't used for much other than calculating inducements.
plasmoid wrote:For me - in a League setting - I wouldn't want a handicapping system that compensates for skill. In a League setting I prefer the better coach to have the advantage. In that sense I prefer a handicapping system that tries to gauge just mechanical strength rather than a combination of mechanical strength and player skill.
I prefer that too. What you, and many others can't wrap your head around is that there is presently no way to do that... and, in fact, that is not what you're doing with your suggestion, either.
So, with that, lets talk about
METRICS.
As you may recall from past discussions, a
metric is an objective, numeric measurement that we use to determine the effects of change in an experimental environment. It's not simply important, it is
utterly mandatory if we want to even flirt with validity.
When someone says their car is better than your car, the first question that should pop into your head is "better how?". That's your instinctive understanding that "better" or "worse" require a metric or they're just shit-talk like that sign that says "Voted #1 pizza in the city!" over every single pizza parlor in town. Voted by whom? The owner of the shop? It's meaningless without the objective metric.
So then, what's your
metric for determining positive or negative effect of your proposed changes in terms of TV being a measure strictly of mechanical advantage and not coaching skill? Are we going to plug in the arbitrary change, have people play 100,000 games, and then see if the effect of TV difference on win rates has decreased? At what point will we say that TV is now perfect... when there's no difference in the win rates based on TV? That means that every modification of any sort to your valuing system requires another massive number of games in order to see if you've positively or negatively affected the metric.
If we're also going to accept the idea that some rosters are, or even might be better than others we'd need to examine the inter-roster win rates as a factor of TV difference and see the differences decrease over time.. since the differences in the rosters are really differences in TV's application to the combinations of stats and skills possessed and available to them. Any intra-roster comparisons jack our required dataset up by a factor of... hell, 100? maybe more. It's rough enough that datasets like FUMBBL's B over a year were barely large enough to examine some rosters... and weren't enough to examine play at higher TV levels.
That brings us to the second (or primary, really)
massive flaw in what you're trying to do: if your aim is to price each skill appropriately then each skill is a variable itself. The only way to test the effect of a single variable is to control all the others and then deliberately adjust it... and the only way to know that we've made an "appropriate" adjustment is if all the other variables are likewise appropriately priced... otherwise we may simply be over or underpricing one skill which is compensating for the over or underpricing of another skill.
Ultimately even if,
across the span of decades of trial and error, you found a pricing setup that eliminated TV difference as a predictor you'd still be unable to say that individual skills were priced appropriately, just that the collective cost of popular skills seems to be high enough to make the overpriced inducements better available as a means of bridging the gap between teams of different TV.
plasmoid wrote: And basically there are 3 obvious ways of trying to make all skill choices more equal.
1) Try to buff, nerf and merge skills, to actually make them equally valuable.
2) Assign different values to skills, and let coaches select a set value on each skill-up (basically letting coaches choose which skills to merge)
3) Assign different values to skills, hoping that team management will take care of the rest.
#3 is to my mind the least intrusive. Which of course raises the question whether it would do enough.
You can do #3 by doing nothing at all. It's what is already in place - team management is already what is compensating for the varied values of skills. Changing the cost of skills arbitrarily is... nothing but changing the cost of skill arbitrarily under #3.
Let me give you a #4 that is obvious to the stats folks: create a frequency distribution of skill selections being made within a given environment (preferably large-scale) and then increase the cost of any skill that falls 2 standard deviations above the norm, and decrease the cost of any skill that falls 2 standard deviations below the norm. Then your price changes are not arbitrary, they're based on team management - meaning they're based on the collective "wisdom" of all blood bowl players regarding the value of certain skills. Repeat the process until there are no skills in the +2z category, and make 5k the minimum cost for any skill.
plasmoid wrote:Those teams will be weaker no matter what. They're designed that way.
TV difference is also "designed that way" since it is primarily about inducement calculations and the designers made inducements overpriced. If you're trying to "fix" TV then you're flat out saying you don't care how the game was designed, you think it'd be better if it were designed differently. When you go that route you don't get to use "but it was designed that way" as an excuse for directly relevant issues.
Ultimately, your proposed system can have its "#1 pizza parlor in all of blood bowl" sign, but until you work out the logistics of applying an objective metric and creating internal validity within the system, that's about all it will ever be able to boast, even if you could secure the participation of the entire blood bowl world across decades.
