Staticness and text sims

[Honolulu Blue]

I'm not sure if "staticness" is a true word, but you'll get the gist of it if you read the rest of this post. Don't worry; it's not too long.

I suspect this isn't a problem for most sports text sim players, some of whom don't go as far or as fast as I do, and some of whom can sim 100 years or more with no problems.

I can't do it. I tend to lose interest past the 20 year mark. There are at least two reasons for this:

1) By the time I've put that much time in the game, I've figured out most of the AI flaws and gaps, and I either exploit them for all they're worth or try a new career with house rules so that I can't. Believe it or not, this is the smaller of the two problems.

2) The bigger problem is that with some limited exceptions, the game is played the same 10, 20, 50, 100, or 500 years from now. Only the players are different.

This starts to become a problem about 20 years into a game. Think back to 1983, for those of you old enough to remember. Has baseball changed since then? Sure. The salaries are bigger, the power numbers are higher, and so are the ERAs. This is partly due to personnel changes, but mostly due to changes in coaching philosophy, workout tactics, and rules. In pro football, the changes have been subtler but still there - yards per pass is down, as are interceptions, team yards rushing per game is down, and also touchbacks. Again, some of it is due to players, but much more of it is due to changes in coaching and rules. Basketball has changed. You'll have to ask my evil twin about hockey, but it has probably changed too. I don't think soccer has changed much but it takes forever to sim out 20 years using CM anyway.

50 years ago, the major sports were even more radically different.

The question is, will football or baseball be the same 20 years from now as it is today? I would argue not. What will change? I don't know, and to guess would merely be speculation. But it's a far better bet to assume that it will be different than that it will be the same.

And this is what most of the major sports career text sims miss, IMO. They assume the game will be the same forever when it is obvious that they change and evolve - and along with it, their record books. Eventually I think the games will reach the next level and either allowing the user the option of globally changing the game and its assumptions, or imposing different rules. But we're not there yet, and it creates problems for me. And probably just me.

[corbes]

You should find and play an old copy of Baseball Mogul. Within 20 years, you'll be into a dead-ball era where every pitcher has an ERA under 1.00 and the best hitter in the league is hitting .180!!!

I think the trouble has been balancing a game that will change with time, and not wobble totally out of control.

[fantastic flying froggies]

Didn't OOTP use to have different eras available ?
That might be a way around this problem... Move from a deadball era to modern than to small ball than to...

[Ben E Lou]

Originally posted by corbes
I think the trouble has been balancing a game that will change with time, and not wobble totally out of control.

Yup. As mentioned above, OOTP does this, but it is a manual, user-selected change in eras. Buc and I (and a few others) had this discussion recently in another thread somewhere, and I had this discussion with Arlie about TPF as well during the beta process.

I would describe what we're looking for as a "dynamic statistics model." In other words, one that could smoothly, and on its own, move from a "balanced" era to a "short passing" era to a "running game and stretch the field passing" era to a "smash-mouth" era to a "run and shoot" era back to a "stretch the field" era, etc. etc. etc.

It seems that this would be a very ambitious undertaking, but one that would be appropriate for perhaps FOF or OOTP to take a shot at, since both are currently 5th-generation games.

[HornedFrog Purple]

I guess I am in the other camp. I enjoy the consistancy but don't mind seeing the occasional burp in a random year or two once in a while. In fact that is really all I want a sim engine to be is consistant even though it might be "off" the real-life average.

Edit: for instance, I believe FOF5 is too pass-heavy but as long as I know that I can adjust because it stays that way for the most part with occasional burps.

[Fido]

This is an interresting topic. Mainly because it may be doable with FOF 2004 as it exists now. There are basically two ways you can implement this dynamic nature.

First would be to make the coach AI much more robust. It would need to be able to adapt to what is working for other teams, but at the same time still be willing to rty new things (trend setters). This may be too much for a text sim game to accomplish.

The second (and doable) way would be to modify the talent pool. The AI seems to do a pretty decent job of working with what it has, so why not take an existing draft file generator and tweak it so that it created players geared towards one aspect of the game or another. It should be pretty easy to set something up that would, say, cycle through decades of pocket passers transitioning to scramblers, power running backs to agile ones, bug stroing DL to smaller more agile ones. I'm really not sure how the game would handle it, but it may be worthwhile to generate 20 yrs or so of cyclical drafts like these then run them into the league to see how the computer controlled teams' stats and tendencies change over that time.

[Honolulu Blue]

Originally posted by SkyDog
I would describe what we're looking for as a "dynamic statistics model." In other words, one that could smoothly, and on its own, move from a "balanced" era to a "short passing" era to a "running game and stretch the field passing" era to a "smash-mouth" era to a "run and shoot" era back to a "stretch the field" era, etc. etc. etc.

It seems that this would be a very ambitious undertaking, but one that would be appropriate for perhaps FOF or OOTP to take a shot at, since both are currently 5th-generation games.

Well, that's part of it. As a human player, I can run any kind of crazy fool offense I want, but it would only have a small effect on the stats. None of the AI players will try to imitate me, even if I'm incredibly successful. The philosophy in the major sports seems to be a version of follow the leader.

But probably the easier way to implement something like this is to imagine a rule change and its effect. What would happen if, in the NFL, all chucking of receivers beyond the line of scrimmage was made illegal? How about if they used the college version of overtime? Narrowed the uprights? Moved kickoffs back to the 25? It's possible to imagine such things and many others - even if none of them, taken individually, is likely to occur - and it's possible to guess about their effects.

Jim took some steps toward this with FOF Classic with his ideas on expansion and the virtual reality seats. But lately he's backed away from it, as if to say, "Aw, who knows about the future anyway? I predict the future will be exactly the same as the past, only with different players."

Someone who can make a career text sim with a dynamic view of the future that's different every time you play is likely to get a disproportionate share of my money.

[Honolulu Blue]

Originally posted by Fido
First would be to make the coach AI much more robust. It would need to be able to adapt to what is working for other teams, but at the same time still be willing to rty new things (trend setters). This may be too much for a text sim game to accomplish.

Coaches do this to some extent already on the game level - they say things like "That play looks familiar to me".

But to do this on the macro level is much, much tougher and would involve something like true intelligence. To say, "That team over there won the Front Office Bowl with a QB that has mediocre ratings other than short passing and scrambling. Perhaps I should get one of those guys and do the same thing." is something human players do all the time, but computer players never do. I think it's part of the next level in career text sims - rather than replaying variations of the same season over and over again, have some computer players introduce a new way of doing things that significantly (if not radically) change the game. It may take awhile, but I think we'll eventually get there.

[Ben E Lou]

Originally posted by Honolulu Blue
But probably the easier way to implement something like this is to imagine a rule change and its effect. What would happen if, in the NFL, all chucking of receivers beyond the line of scrimmage was made illegal? How about if they used the college version of overtime? Narrowed the uprights? Moved kickoffs back to the 25? It's possible to imagine such things and many others - even if none of them, taken individually, is likely to occur - and it's possible to guess about their effects.

I like this idea! It could be a pretty good compromise. Have a list of 10-20 different potential "rule changes" that could take place in the game at times, and each one could potentially have an effect on both statistics, and on what kind of players you'd want on your roster. It doesn't seem like it would be terribly difficult to "prepare" the AI for roster-building pending on rule changes (If chucking receivers beyond line of scrimmage is illegal, then stop including bump-and-run rating in the algorithm that values players, for example.) Each rule change would also have an accompanying statistics change as well, by a percentage. This would allow for changes in statistics output with (I would think), MUCH less fear of things wobbling out of control.

[cthomer5000]

Originally posted by Honolulu Blue
Jim took some steps toward this with FOF Classic with his ideas on expansion and the virtual reality seats. But lately he's backed away from it, as if to say, "Aw, who knows about the future anyway? I predict the future will be exactly the same as the past, only with different players."

Someone who can make a career text sim with a dynamic view of the future that's different every time you play is likely to get a disproportionate share of my money.

I'm with you in that I would love to see him add more of this into future versions of the game. I always loved getting those emails about the new SparkleTurf BioGrass stadiums, virtual reality boxes, etc..

I think having a list of potential rule changes, and then coming up every so often would be nice. Then you could just have the option of turning those on or off, so those who didn't want changes coming into the game.

[skrath]

I did this in my basketball sim. After about 30 years, sometimes the Board of Directors for the league will vote for random things, such as "Setting the 3pt line 2 feet further back" or "Changing the Foul Rule" or "Changing the FT rule", etc. It also generates, depending on the change of rules, different players - i.e. for 3pt line being moved back, shooters are higher valued in the scouts eyes.

It works ok but I can see a few people saying it's a "corruption" of the sport. I guess it just needs to be an option where people can turn it off if they don't want an alternate reality, as cthomer just said.

[Icy]

Well as in mostly of things about text sims, the CM series take the lead. Since the first versions, the computer GM's "learn" from you and how to beat you, this means thay change formations etc. This finally ends on a changing universe. Of course it's not perfect and lost of times it plays players out of their best positions etc. The AI is the biggest problem at any game, that's why we enjoy online leagues as we can play vs human brains. The only problem with online leagues is that you cant play when you want as you can when you play solo.

[Ben E Lou]

Further thoughts, based on just a couple of the potential rule changes mentioned here:

All chucking of receivers beyond the line of scrimmage becomes illegal.
(Devalue bump-and-run coverage. Increase probability of routes being run properly by x%. Increase probability of AI choosing to pass by y%. Increase AI valuing of route running by z%)

Results: Increased passing attempts, yardage, completion percentage. Decreased rushing attempts.

Narrow the uprights.
(Increase value of kicking accuracy by y%. Increase possibility of punting or going for it on fourth down.)

Results: Fewer FG attempts. More going for it on fourth down. Lower FG percentage.


There could be a nice handful of these rule changes, with a chance that any given one could occur, but not too often--maybe no more than one every 10 years or so. Each one would have a few resultant AI adjustments and stats changes.

[cthomer5000]

Originally posted by SkyDog
There could be a nice handful of these rule changes, with a chance that any given one could occur, but not too often--maybe no more than one every 10 years or so. Each one would have a few resultant AI adjustments and stats changes.

Exactly how I would like to see it implemented. Maybe the rule changes could be every 10 years or so, with a variance of 3-4 years in either direction.

[Huckleberry]

I see no problem with that feature being added as long as it can be turned off when beginning a game by choice. That way you can still play without it there or wait until you're convinced that it's reasonable.

A step further would be to allow the player to select what rule change they want to implement from a supplied list.

[Fido]

Originally posted by SkyDog
Further thoughts, based on just a couple of the potential rule changes mentioned here:

All chucking of receivers beyond the line of scrimmage becomes illegal.
(Devalue bump-and-run coverage. Increase probability of routes being run properly by x%. Increase probability of AI choosing to pass by y%. Increase AI valuing of route running by z%)

Results: Increased passing attempts, yardage, completion percentage. Decreased rushing attempts.

Narrow the uprights.
(Increase value of kicking accuracy by y%. Increase possibility of punting or going for it on fourth down.)

Results: Fewer FG attempts. More going for it on fourth down. Lower FG percentage.


There could be a nice handful of these rule changes, with a chance that any given one could occur, but not too often--maybe no more than one every 10 years or so. Each one would have a few resultant AI adjustments and stats changes.

These changes are in the very minor category, and the same thing could be accomplished by simply modifying the draft files. You can simulate the narrower goal posts by modifying the kicker's accuracy and no bump and run by reducing (drastically) DB's B&R potential.

The real issue is with major changes. Two point conversions, and replay come to mind right away. The AI algorithms would need to be written to handle all such new rules, even if the rules are never enacted. What if one of the rules was that you could go for a four point conversion by running a play from the 10 yard line. The computer coaches would need to have tables built in telling them when it is appropriate to try for it. That brings up the issue of the human player having to get up to speed with the changes.

I guess the issue I have is with HB's assertion that changes in the games are due mostly to rules changes. I think that the game changes much more based on talent of players. Look at how QBs like Vick are changing the game. That is completely based on his talents, not on any rules changes. In baeball, power numbers are up and ERAs down because the batting talent is growing more rapidly than the pitching talent. To me, the whole dynamic nature of things is based on changes in talent and each team's ability to react to it.

[Ben E Lou]

Originally posted by Fido
I guess the issue I have is with HB's assertion that changes in the games are due mostly to rules changes. I think that the game changes much more based on talent of players. Look at how QBs like Vick are changing the game. That is completely based on his talents, not on any rules changes. In baeball, power numbers are up and ERAs down because the batting talent is growing more rapidly than the pitching talent. To me, the whole dynamic nature of things is based on changes in talent and each team's ability to react to it.

I agree with this in principle. It is just that I'm operating under the assumption that creating a stable, yet dynamic, player talent pool that would require the AI to make subtle changes over time. (More Vick-type QB's get created, then the AI would need to slowly change as they become the norm.) On the other hand, a set of rule changes that would cause more/less passing, more/less running, more/less FG attempts could each have a defined AI response that would ensure that the human player wouldn't gain a greater advantage than he already has.

[JonInMiddleGA]

Although I'm not really sold on random rule changes as a feature, another way to handle the implementation is to leave it on but have an "owner vote" segment where your "Nay" vote would serve as a (actual but unofficial) veto, but a "Yea" vote wouldn't guarantee the implementation either.

In other words, if you vote "yes" then it might happen.
If you vote "no", then it won't happen.

[Ben E Lou]

Originally posted by JonInMiddleGA
Although I'm not really sold on random rule changes as a feature, another way to handle the implementation is to leave it on but have an "owner vote" segment where your "Nay" vote would serve as a (actual but unofficial) veto, but a "Yea" vote wouldn't guarantee the implementation either.

In other words, if you vote "yes" then it might happen.
If you vote "no", then it won't happen.

I'd DEFINITELY say that a potential rule-change option should be just that, optional. There would be times that I myself wouldn't want to use it. One thing I LIKE about a static model is that 1,500 rushing yards in 2003 means the same thing as 1,500 rushing yards in 2023. There would be times that I wanted to see that, but there'd also be times that I'd want things to change over time.

[Fido]

Originally posted by SkyDog
I agree with this in principle. It is just that I'm operating under the assumption that creating a stable, yet dynamic, player talent pool that would require the AI to make subtle changes over time. (More Vick-type QB's get created, then the AI would need to slowly change as they become the norm.) On the other hand, a set of rule changes that would cause more/less passing, more/less running, more/less FG attempts could each have a defined AI response that would ensure that the human player wouldn't gain a greater advantage than he already has.

Very true. But the issus is still that Jim would have to come up with a list of every possible rulr change, and code the AI to handle every possible combination of rules changes. Couple that with the fact that he woudl need to code it so that the AI gradually gets used to the rules and it becomes quite a time investment. I, for one, would MUCH rather see him spend his time of other things (customizeable league sizes, and game plan encryption for MP leagues for example).

[judicial clerk]

The expansion, virtuality seats, etc. were cool and did add something to the game.

[John Galt]

Originally posted by SkyDog
I'd DEFINITELY say that a potential rule-change option should be just that, optional. There would be times that I myself wouldn't want to use it. One thing I LIKE about a static model is that 1,500 rushing yards in 2003 means the same thing as 1,500 rushing yards in 2023. There would be times that I wanted to see that, but there'd also be times that I'd want things to change over time.

Damn conservatives are always trying to regulate and control changes.

[Ben E Lou]

{attaches mistletoe with John Galt's name on it to coattail}

[Ben E Lou]

[Godzilla Blitz]

What a great idea, HB!

I think that rule changes are an obvious way that the game changes, and would be fun to implement in a text-sim. In football, for example, the rule change in football allowing free substitutions had a huge impact on the way the game is played now: the rule engendered a player specialization that has resulted in huge players being able to play the game. Back when substitutions were limited, a 300-lb. lineman would be a liability because he'd be exhausted 20 minutes into the game after playing both sides of the ball. Football shape and size changed, hash marks were brought into the game, etc. The potential is endless. How you would encorporate all this into a game, though, seems somewhat difficult. It seems that text-sim designers have a hard enough time trying to get realistic stats with a static sim.

Another way to approach this, I would think, would be to simply use the existing game structure, and use event triggers to similate a change in the quality of athletes in the player pool, or other small changes that could be easy to implement. For example, you could have event triggers like:

1. Drugs. Think steroids, baseball, power, sport's image. It would be neat if a game could portray a drug policy that was either lenient (baseball) or strict (football), and then portray the impact of those policies on the sport. For example, in baseball, when you get to some point in the sim, the game could trigger an event where the league votes on a drug policy. If it's lenient, then players will gradually show increased power numbers over the next decade, but the league, and you as a GM, will have more events where your players are accused of/suffer the effects of using drugs, and attendance could drop as the sport's image falters. This would be something the existing AI routines could probably handle fairly easily.

2. Biotech. One of the exciting things on the horizon is anti-aging drugs. Right now it's still a thing of science fiction, but what if in a text-sim, when you arrive at some point in the mid-twenty-first century, athletes aging routines slow down by xx%? Instead of an athlete fading away after 15 years, they would reach their peak at the same time, but an average career would last 30 years or more. The rate of aging reduction could be less, or more, depending on a random factor. This would obviously take some more coding to handle, because if you double the length of an athlete's career, you'll have double the number of eligible players in the player pool, but to a certain extent, this would reflect reality as well. What kind of decisions would a GM have to make if star quarterbacks could have a career of 30 years or more? As an aside, this, IMHO, is one that to some degree will become reality within 50 years.

3. Another element to bring into play would be equipment changes or medical techniques that impact injuries. For example, simple ones in football could be helmets that completely eliminate the possibility of concussions. Or improved ACL surgery techniques that make two-week, full recoveries possible. You could even tweak this a bit more, and create choices in the game. For example, the non-concussion helmets could eliminate the possibility of concussions, but be 20% heavier, resulting in players that use them having slightly worse reaction times. Or the improved ACL surgery could have a 25% risk of complete failure, immediately eliminating the player's chance of a return.

The possibilities are endless!

[Huckleberry]

What would widening of the hash marks change?

FG accuracy would go down on field goals attempted after OOB plays and plays that finished towards the hash marks or outside of them. This would have a greater effect percentagewise on shorter field goals.

The percentage of long runs on plays run to the short side would go down. Would theoretically go up to the wide side.

Effectiveness of the rollout passer would presumably go up, particularly when the wide side of the field is to his throwing hand.

Effectiveness of a punter to punt the ball OOB deep in the opponents' territory would go up.

At first I was just thinking about that because I was curious. Now I see what a chore adding this to a game would be.

[Honolulu Blue]

Fido makes some good points, such as these:

These changes are in the very minor category, and the same thing could be accomplished by simply modifying the draft files. You can simulate the narrower goal posts by modifying the kicker's accuracy and no bump and run by reducing (drastically) DB's B&R potential.

Actually, with rule changes like those, you'd have to not only modify the drafted players, you'd have to modify all the players in the universe, since they would be affected the same way also.

Modifying the draft files would work for more subtle changes over time, like bigger offensive linemen, bigger QBs, and bigger WRs.

I guess the issue I have is with HB's assertion that changes in the games are due mostly to rules changes. I think that the game changes much more based on talent of players. Look at how QBs like Vick are changing the game. That is completely based on his talents, not on any rules changes. In baeball, power numbers are up and ERAs down because the batting talent is growing more rapidly than the pitching talent. To me, the whole dynamic nature of things is based on changes in talent and each team's ability to react to it.

I stand by my assertation.

Let's look at what I think has been the biggest change in football over the past 20 years - the gradual shortening of passes. Are QBs of today less capable of throwing it deep than the QBs of the recent past? I doubt it. The main reason passes are getting shorter is that coaches are calling more short passes in proportion to long passes.

With kickoff distance, there's a joint effect - the kickers kick the ball farther, followed by rule changes to neutralize this.

In regards to power hitters vs. pitchers, I'll agree with you to some extent, but I think the other factors at work (smaller ballparks, modified hitting techniques, more resilient baseball) overwhelm it.

It's not that I don't think the evolution of players over time isn't important, and not that I don't think that one or a few players can influence the path of a team or even a whole league. But I think, overall, that coaching moves and rule changes are more important.