Monday, March 04, 2013

Gridiron Solitaire #45: Finally (Finally)

Yeah, it took a few more days, actually.

First, I wound up removing all text that indicated card ratings. It just didn't look right, and it was marring the layout, which I was very pleased with otherwise. However, that introduced another issue: without text, how could I communicate the value of the card that was +/-2, depending on the outcome roll?

New look:


Instead of numbers, value is now shown via stars at the top of the card. Much cleaner, and I have a universally positive reaction to stars. I wish my life had more of them, the kind that the elementary school teachers used to peel and stick on papers for good grades.

Also, if you look at the middle card, you'll see that biohazard symbol to the left. That warns you that the card can go negative as well as positive (and I'm adding the little help referee with a dialogue box to explain that).

Next, Fredrik finished the final accountant image for the reveal. Here's how he looks:


He's watching the draft on television, with the adding machine smoking in the background. If you enlarge that image, you'll see a terrific amount of detail in the accountant's face (no surprise, because Fredrik is so talented).

I also rewrote the CPU AI for the offseason last week. The original AI was solid enough to work, and probabilities were shaded for teams to improve their worst ratings over time, but lots of what happened was based on the randomness of the rolls.

I wasn't happy with how teams evolved over time, because I didn't feel like enough teams had distinctive styles. So I rewrote the AI to have them use the same system as the human player does now--they get a budget, and they buy cards based on their budget. Cards can go bust, just as for the human player, and their priorities can change based on how successful they are as a team.

Here's how CPU priorities are determined, and it's really simple. There are basically two strategies: "same" and "new". With the "same" strategy, teams will improve their best and worst ratings first. With a "new" strategy, they'll improve their two worst ratings first.

How does a team determine its strategy? The chances of a team defecting to the "new" strategy rise with each loss, so a team with 12 losses has a much higher chance of defecting than a team with 3, for example.

There's more to it than that, but it's a basic outline of what's going on now. And it should mean that teams develop distinctive styles, as well as more teams with at least one very high rating, which makes it more like teams in the real world.

Now that the offseason is really, REALLY done, I've revised the help screens, and I'm starting on the dynamic help for offense and defense this afternoon. I'm slowly working myself into a stronger position as the major changes have taken place. Now I just want to optimize the sub-systems (like the sound engine, where I'm going to make one more pass) and get ready for the second beta.

Site Meter