Posted by verbal_kint on 6/18/2012 2:22:00 PM (view original):
Posted by norbert on 6/18/2012 1:24:00 PM (view original):
I'm pretty open to complaints about the game, but calling me a liar deserves a reply. When did I lie to you about your Nebraska team? If I told you something that I found out later to be wrong, I'm sure I was the one that told you I was wrong. How did the updates make the game worse? What part of it is worse? I have no clue what you are talking about. The FAQ is the FAQ from when I took over the game and the way things work in the engine currently many ratings are averaged together for many different things, but HANDS is weighted more in the average rating for receiving so technically it is most important for catching the ball. The problem is that it is not independently a factor in catching the ball. All I can do is work on making a better game and if you don't think I can, then fine, you haven't really seen anything from me so I get it, but to call me a liar and say I only care about taking people's money is just flat out insulting.
This is an absolutely terrible explanation. Even if you take the average HND of the WR on the field, they're terrible as a whole. Please try again.
...and thus the reason for the next few months of working on the engine. I'm not sure where it's getting lost in translation that I feel the current engine is too random as well. This particular focus on the Hands ratings and my explanation of why they are too random can be broken down into several components. One is that the ratings are diluted by averaging them together so one particularly low rating isn't going to have as much effect as if you looked at it separately, yet in the formula of seeing if a ball is caught, Hands is weighted highest of those ratings, so technically it is more important. But the other component of the discussion is that the results are too random and this is where no matter how much Hands is factored into the chance to catch a ball, the results won't seem to reflect it.
I've talked about this before, but from reviewing how the engine is working in determining results, the problem of randomness lies in the way it determines the results. Basically it calculates some chance (a value between 0.0 and 1.0) and then gets a random number (between 0.0 and 1.0) and if the random number is less than the chance, the result happens. For instance, in the case of a catch (or drop), it will average a bunch of ratings together for the receiver and use a formula to come up with the chance of the catch, like 0.9 (not actual values). Let's say the player's ratings are worse so the chance drops to 0.87. You might say a 3% drop in the chance is significant, but since we are only checking this a limited number of times, the difference between 0.87 and 0.9 is not that significant. So someone with the 0.87 chance might actually perform better than the player with the 0.9 chance.
What I am working on now is to change the engine to replace these types of checks with a more deterministic result check based on team and player match ups, and in some cases looking at individual players skills. There will still have to be random checks. A player thrown to 10 times won't either drop every pass or catch every pass. But you should see a player with lower Hands drop more balls than a player with higher Hands.
It will be difficult to pretty much rewrite the engine and get games playing through with the changes and not expect some problems, but that's why we are going to have an open beta. The other difficulty is in trying to update the pbp so that you can see some of these effects and better evaluate your team.