(This is a guest post from our very own Trumwill!)
I present to you a Tecmo Super Bowl video of Bo Jackson evading defenders for an entire quarter. Well, the quarters in TCB are five minutes long on a fast clock, but still.
I used to play Tecmo Super Bowl a lot. It was a groundbreaking game, for both good and ill. I played several teams, including the Los Angeles Raiders (Bo’s team) for a few seasons. The Bo Advantage really cannot be overstated. Just hand it to him, and you’re golden.
I never won the Super Bowl with Bo, however. I found it difficult to win with any team because it cheated. Hard. The better you were, the better your opponents would become. It would start injuring your players. You’d start fumbling incessantly. The opposing players would start knowing your playcalls. And they’d suddenly become really, really fast. Passes would be turned into interceptions. In the case of Bo Jackson, I threw the ball all of nine times all season. Eight of those times the passes were intercepted. And then Bo would get hurt.
And eventually you would have to play either the New York Giants or Buffalo Bills some other impenetrable team.
There was a time when I would start the season just throwing games. Trying to lose or cut the magin of victory. But the early season – before it ramps up – is so incredibly easy that it’s simply no fun.
I finally gamed the system by simulating the first eight games and then choosing the worst team with the most potential. It turned out that was the Philadelphia Eagles, at 2-6. Randall Cunningham (known as “QB Eagles” in the game because he was one of the few players that didn’t sign over his name rights) was really all I needed.
Not that I could use Randall Cunningham, mind you. Because as soon as it got wind of how good it was, Cunningham would be hurt. So I went ahead and started Jim McMahon, the capable backup. So if someone was going to get hurt, it was going to be him (they never hurt all of the occupants of any position). They took out my runningback for a game. Which was a good reminder that I needed to play only backups.
Which I did, finishing out the season at 10-6. They must have known was I was trying to do, however, because they never injured any of my players once I put the backups in. In the second half of the Super Bowl, I finally decided that it was “now or never” with Cunningham and the rest of the started. Randall Cunningham was hurt within four plays and a runningback followed.
But I still won my first and only Super Bowl. I played the Eagles again for another season. I used the same tricks, but I still won too many games and couldn’t overcome the Giants at superspeed and lost before making it to the big game.
I LOVED the Kansas City Chiefs team on that game. The offense? Aside from Christian Okoye, Barry Word, and Stephan Paige, there wasn’t much going on. Steve DeBurg’s mastery of play-action just didn’t translate to the pixellated screen, and the only other real receiver I had was JJ Birden (I don’t remember the #3). The trick on offense was to understand that the computer would always leave one receiver open somewhere, and even when covered, you could usually hit a good receiver if you timed the pass perfectly to when he made his cut.
But the defense! Oh, the defense. Percy Snow. Derrick Thomas. Neil Smith. Derron Cherry. This was a team, both in real life and in the game, where we fans suffered through the offensive possessions because we wanted that magnificent defense on the field. Derrick Thomas would end up with some ridiculous sack total (something like 30 or 40), and I could usually get the rest of the D Line into the low teens, too.
We spent a lot–A LOT–of time playing this game. I could beat the computer any way you wanted me to, with pretty much any team (the real challenge was winning the Super Bowl with that dismal Dallas team, or that horrible New England team). Where I failed was in playing another person. I was really good at identifying the patterns the computer played and could exploit them on offense or defense. Where I failed, though, was in playing against people. I absolutely SUCKED at playing against people! We played a season wherein we all took a team in the same division and I think I won a single game against a human opponent. I barely won a game once when my buddy took a really crappy team and I took a top team. I couldn’t do anything but throw up desperation bombs and I kept completing these ridiculous passes for long yardage and touchdowns.
Awesome game, though. One of my all-time favorites.
I did a jaunt with the Patriots. Same thing happened with them as happened with everybody. Great at first but eventually I had to play the Bills.
I never played Techmo, but Madden ’95, you could cornerback blitz with Dion Sanders almost every play and get a sack against the quarterback. You could get to the QB usually before he could hand the ball off, even.
We put Neon Dion on the Verboten list, after that.
The better you were, the better your opponents would become.
I despise that trick in games. I like playing Real Racing on our IPad, but it does the same thing. Opponents are too easy to overtake at first, but once you’ve passed everybody, a couple of them suddenly become much faster. I sometimes find that an early lead is impossible to hang onto, but if I hang out in second or third position and only go for the lead in the last lap or two it’s easy to win. The problem is that because you’re behind it makes those others cars slower, and it can be really damned hard to stay behind them. They’ll make ridiculously slow turns, and it’s difficult to not run right up on them and either pass or crash.
Why the hell do game designers do that?
Why the hell do game designers do that?
There is a real answer to this question. It has to do with AI and how difficult it is to write even a mediocre one that is capable of doing something like “drive around the track, as fast as you can, avoiding the obstacles”.
It’s easier to use a rubber band.
Programmers just need to figure out how to code this.
Long ago, in a fit of conscience, I swore I won’t implement a homing algorithm or work on a weapons system. I’ve stayed true to that promise. AI (or more properly, homing algorithms) solved the Steering Problem very early: radars gave us excellent ranging data and simple trigonometry gave us the rest, even for supersonic vehicles. The old Hawk Missile system hit everything it was ever fired at.
The Mars Rovers steer autonomously: they are perfectly capable of navigating through a rock field. It’s too far to Mars for ground control to actually steer them: we give them an objective and they’ve got enough onboard sensors to work out how to get there.
In answer to James Hanley’s question, each of the cars on the track is operating on its own feedback loop in concert with a larger situational matrix. In plain English, when “you get your game on”, the AI gets its own game on in response. The situational matrix recognises a challenge. There’s a correspondence in the real world of racing: the car in front is burning more fuel: being out in front, early in a race might put lead lap points on the board but it won’t necessarily put you in the winner’s circle. It will, however, put you in the pits earlier.
A mechanistic trader pitted against a human trader can always respond to data faster. It may not always make the smartest move, though. It will only parse the data and respond on the basis of its rulesets. Mechanistic trading is almost always supervised by a human trader: AI can never model every aspect of trading.
I haven’t written anything like a classic video game since the Atari 800 and that wasn’t a real game which went to market. I have, however, worked on AI components which know how to “speed up” and “slow down” on a situational basis. In many circumstances, in fact now that I think about it, in all circumstances, the AI knows how to “scram”, retreat to a safe position when it’s transiting out of known safe parameters.
Because they like trolling you. I knew a guy who wrote an AI that would consistently make the player lose. the catch was, that the closer you got to victory, the more improbable the loss would be. “Player mysteriously transported into pool of fire, mere inches from victory”…
That was a BAD AI. No. Fun.
(I think sports games tend to suck because game designers don’t tend to like sports. “You’re the only American, you go write the basketball game — you know the rules, right?” tends to make bad games).
I saw a good “visual-novel-like” game whose core mechanic was hanafuda. I’d recommend it, but the english translation is HORRID.
This was a great great game, though I never played a season. I didn’t own it, but one of my friends did, so I just played against him.
I think it was one of the Techmo games where I realized that I could dive away from a defender anytime they attempted a tackle. I’d get up sooner than the defender and go for a big gain. My friend did not appreciate that strategy.