More Blood Bowl AI

I was going to write up this long AI critique today but my Dr. has prescribed happy pills for my back so WOO HOO. Go Happy Pills. However, the BB addicts on the game's official forum have done the work for me so in that honor bound blog tradition I am going to totally steal that info and paste it here.

This is in very basic terms what is wrong with the Blood Bowl AI whether you play on Easy, Medium or Hard. It matters not.

This from a fella called Bitterman:

Speaking as a games developer, writing a good AI isn't easy. Especially for a game as complex as Blood Bowl. Some of the suggestions above are great advice for human players but a computer will really struggle to understand it. But there are a few easy wins that absolutely requirement improvement for the game to be at all challenging and interesting in single player.

1. Understand turns 7 and 8. It is completely pointless (and totally ruins any illusion of intelligence) when a team 2-1 down at the end of turn 14, with the ball, doesn't advance anyone on turn 15, making it impossible to score on turn 16. This is absolutely critical and 100% required that this behaviour gets fixed.

2. Differentiate more between team tactics. It is completely pointless sending Dwarf Blitzers forward on the wings in the same way as Wood Elf Catchers. Effectively it's like the Blitzers aren't involved - so you're playing against 9 men - I've yet to concede against Dwarfs, or win by less than 3-0 (usually 4-0). While on the subject - if a Dwarf team is kicking, it's not always a good idea to field the Deathroller - I've never had to face a Deathroller for more than three turns before I score and it's sent off.

3. Differentiate more between kicking and receiving. I'm not sure I ever remember more than three players being set up on the LOS, with two in each wide zone and a line behind the front. It's so predictable - and the same whether kicking or receiving. Why not set up seven Dwarfs on the front line sometimes? Or four Wood Elf catchers and both Wardancers on one flank when receiving, to overload it? Every drive is the same.

4. Team progression is broken. Goblins with Break Tackle are just the start of it. There's no sense that the team develops as a unit - players are given skills apparently at random, in isolation. This is probably the most complex fix because there's a lot to it (strategy, trying to predict future tactics, rather than just tactics in isolation!) but a few simple rules would make big differences.

It's a good game, it's fun - I mean, it's Blood Bowl - but the AI is very weak in the above areas. It'll never be as strong as a truly skilled human but on the "hard" setting, it should not do things that are just stupid - but it consistently does, usually leaving it incapable of scoring when it needs to equalise or win - no human would make such a basic mistake.

He's on point with all of that.

However it goes deeper.

In order for the AI to be a threat -- not to play a brilliant game but just to pose a threat to a good BB player it needs to stop with the excessive dodging and the ridiculous amount of times it tries to "Go For It." -- Going For It is the game's defacto sprint -- you can move 2 spaces more than your Move Allowance but you need to roll a D6 each time you move and if you roll a 1 you fall and your turn ends.

The AI Goes For It every turn and with multiple players. It is TRYING to set up a "cage" -- this is a BB term where the ball carrier is placed in the middle of a bunch of teammates who shield (cage) him from enemies. A common tactics with teams like the Orcs, Dwarfs, and Chaos. Of course the AI does this with every race...

But it is SO intent on this cage idea that it Goes For It like crazy in order to set it up and invariably someone tumbles, screwing up the turn. They also like to dodge...with Dwarfs. A lot. That's 10 ways of stupid.

Finally -- MOVE THE BALL. The AI will only move the ball when it has a wide open path. Of course a good BB coach won't allow that unless the dice just flat out betray him. A good offensive coach has to plan blocks, and yes, take a few risks with passes, blocks, nullifying tackle zones, etc. The AI is terrified of your players with its ball carrier. And will not move until the road is wide open. This is why my games are almost all 2-0; 3-0; 4-0. The AI will score every few games based on abysmal die rolls but it doesn't get credit for that in my book.

So yes, the AI has no sense of time, no sense of the score, no sense of its star player's skills and how best to use them or how to line them up, no sense of game planning based on race, no sense of basic friggin' percentages when it comes to dodging, goes for it too much, and has no idea how to mount a basic offensive attack.

Other than that, it's awesome.

Look, this is still Blood Bowl and playing the game online is about as good as gaming gets for me. The rules of the boardgame are mostly in effect and are brilliant. It's a wonderful game. But Cyanide needs to fix the AI like...soon. Otherwise when all of the new players who are learning how to play the game start bitch slapping the AI (it will happen, believe me) boredom is going to set in.

I just hope for your sake that it takes a long time for that to happen so you feel like you got your $50 worth.

Shadows Over Camelot

This weekend my buddy hosted the final phase of a monthly AD&D campaign in which I've been participating for the past eight months or so. It's been my first exposure to the pen and paper RPG world after a lifetime of playing RPGs on the PC and at some point, on this blog, I might dig into that experience a little deeper. (It was a jolly good time.) But for this post it serves as a segue to the boardgame Shadows Over Camelot (from Days of Wonder), which one of the guys brought over to play after the AD&D campaign wrapped up.

I'm not nearly the board gamer Bill is, but I'm on that road as I've become a huge fan of games like Railroad Tycoon and Age of Empires III. And I've enjoyed playing some other titles to which I've been exposed over the past two years, including Shogun, Powergrid, Civilization, Ra, and one of the Lord of the Rings variants. (I forget the specifics, but it involves each player taking on a member of the Fellowship in a quest to destroy the ring.) I'm not sure I liked Shadows nearly as much as I enjoy playing Railroad Tycoon, but it is, without question, a fun game.

In Shadows Over Camelot each player becomes a specific Knight of the Round table. (Lancelot is absent, but you can select King Arthur.) The board is centered on an under siege Camelot and its Round Table. Around it are various quest areas where your knight can go to accomplish various goals like fending off a Saxon invasion, defeating a black knight, finding Excalibur, or going on the quest for the Holy Grail. Since each player is a Round Table knight, you each set off on quests to tackle either single-handedly or -in some cases- cooperatively. But you're all working towards the same goal, which is the defense of Camelot.

Each turn you must enact both an evil and a heroic action. Evil actions include turning over an evil card (which has some less than savory ramification), placing a siege engine outside Camelot (if 12 engines surround the castle, the knights lose) or losing a hit point (each Knight starts with four hit points, represented by a six-sided die). Heroic actions involve drawing heroic cards (if the knight is in Camelot), moving to a new quest location or playing hero cards (usually at said quest location). The Hero cards, for example, might involve getting closer to recovering the Grail or Excalibur. Some have special actions, like allowing all the knights to acquire another heroic card regardless of their locale.

Successfully completing quests results in multiple bonuses, but mainly the acquisition of white swords for the Round Table. Failing a quest (and a few other evil events) results in black swords being laid at the table. When all twelve sword slots are filled the game ends. More white swords equals victory, more black swords means defeat.

In this regard, it's pretty standard fare. It's fun, especially for the co-op aspect of it, but nothing special. What really added a great element to the game is the presence of a traitor in your midst. At the start of the game you're given a loyalty card that only you are allowed to see. It states whether your knight is true or a traitor to the cause. If your knight is the traitor it's your job to try and subtly steer the true knights towards defeat, while hiding your true aims for as long as possible. You can try to guess the traitor as part of your heroic action, but if you're wrong a black swords is laid on the Round Table. (You also can't guess until there are at least six siege engines outside Camelot.)

For our particular group (five players in all) this element of the game really took it to another level as we spent nearly an hour all working together, but not quite knowing who to trust until the traitor eventually revealed himself, doing considerable damage to our cause in the process. (For those in the know, he played the Guinevere card, which resulted in the loss of two of our quest battles.) Despite the setback, though, the four true knights (no, I wasn't the traitor) eventually prevailed.

You can never quite tell how you're going to feel about a board game like this from one play. Sometimes the exciting/fun elements of a game like this grow stale quickly. (I suspect if I get to play more Age of Empires III, that it would fall into that category.) But Shadows Over Camelot certainly looks to have that certain something that would make it a fun repeat play and I'm hoping to get another crack at it the next time our little gaming group gets together.