The amazing adventures of RoboPlayer

[JMendes]

Hey, H. :)

I take it your question is what Einstein would call gedanken experiment, literally, experiments in the mind. If so, then it's a damn good question. Let me try to spit out my answers:

Can RoboPlayer play this way?

Two possible meanings:
a) Is the above definition possible? This meaning is irrelevant in the face of the experiment, so I'll just discard it.
b) Is the above definition considered to be 'play'? A bit more interesting, but still, as we've been seeing lately, different people have wildly different definitions of play, so I'll also skip this one.

Can RoboPlayer be a good player this way?  Could he be one if he had a really good program?

Hmm... What's a good player? Sure, if we develop the program to the point that it emulates a real human, then yes. But that breaks out of the experiment box again, doesn't it? And until then, simply, no.

What can a human player do that RoboPlayer can't?  (One answer is "play author/director stance games", I think.  So, what can a human player do in an Actor stance game that RoboPlayer can't?)

You already got all the straight answers to this: take initiative, be proactive, play creatively. I just wanted to note that the answer you advanced is not a very good one, formally speaking. Simply put, there is no reason why the program can't change stances, given it's fed some appropriate parameters.

Is it possible for a GM *not* to railroad RoboPlayer?

Ahh... Now, we're getting to the good stuff. Took me a bit to come up with an answer. Yes, provided that: he can keep oblivious to the fact that it's a program; or he plays strict simulationist and generates all external events in a formulaic manner (almost being a program himself, you might say).

Is it possible for a GM *not* to railroad RoboPlayer if the GM knows RoboPlayer's program and the program contains no ambiguity?  How much of the program can the GM know before this problem arises?

Under these conditions, the first option above disappears and you are stuck with the second one. The answer's still yes. I'll also note that, in this case, RoboPlayer has effectively become part of the event generation program.

Another interesting question pops up: is RoboPlayer useful? I think so. For instance, for setting up dry runs or alpha tests of settings or pre-made adventures.

Good stuff, hyphz.

Cheers,

J.

P.S. Hmm... I just saw your followup and realised that you were going along entirely different lines. Anyway, hope you find this interesting. :)

[Thierry Michel]

I take it your question is what Einstein would call gedanken experiment, literally, experiments in the mind. If so, then it's a damn good question.

My thoughts, exactly, and I was waiting for RoboGM to show up. Ah, well...

[Le Joueur]

What you've defined is the responsive ideal.  This player is totally reactive, not proactive in any way.  I watch humanity sometimes with an 'outside perspective' and one thing I can tell you is they want interaction, not just reaction.  Oh it's easy to find people who get a little too much interruption willing to extol the virtues of a simply reactive partner, but left alone with them, it pales.

I wonder why you assume this.  Nothing in RoboPlayer's system above bars the program from being one that gives proactive actions high scores.  Yes, it only performs actions that have been indicated as possible by the GM, but in a non-authorial game all players must do that, because in these games, unless the GM's described it or it can be reasonably entailed from the GM's description, it isn't in the world.

Perhaps because you characterized him as thus:

[*]Using GM's descriptions and game rules as input, construct a "database" of potential actions and how they relate to each other;

[*]Using some information about a 'character' to play - either internally generated or supplied to it - together with a 'premise', 'social contract', 'group focus' and similiar, generate a "program" of judgment rules that return score values for potential action;

[*]When time comes to act in the game, look at all possible/appropriate actions in the database, judge each action using the rules in the program, and emit the one that produces the highest score.[/list:u]

What I hear is "Using...input, using...information...generated or supplied to...generate rules....  When time comes...look...judge...and emit...."  Everything 'used' must be 'received' first, then acted upon.  That's strictly responding to stimuli, reactive behaviour.

I am surprised by your comment about non-authorial games; I regularly launch into all kinds of mischief that were in no way offered by the gamemaster without once taking Author Stance.  That's what I call 'taking the initiative.'  "Can be reasonably entailed" doesn't sound like something a Roboplayer would pursue.  (Likelihood of unmentioned data being outside of acceptable parameters = high; do not explore.)

The subtle thing at work is that, over the long term, no matter how attentive, obsequious, and sycophant-like, the Roboplayer will slowly give the impression that they just don't care.  Longer and it actually starts seeming insulting.  Everybody needs 'somebody on their side,' but in the complete absence of criticism, the 'disconnectedness' will read patronizingly.

Again, why should it be obsequious and sycophantic?  You can just change the judgment program to return lower scores for these actions.

Ahem, did I say obsequious and sycophantic?  Sorry, I meant helpful and courteous; yeah that's it, helpful and courteous¹.

Even so, given these are the only behaviours (attentive, helpful, and courteous), and in the absence of creative and initiating actions, trust me people will pick up on it as being obsequious and sycophantic.  And that way lays the bane of being surrounded by nothing but 'yes men.'

A little Roboplayer now and then will really 'grease the wheels' of the game, but as a permanent member, I see serious problems.

Fang Langford

¹Wow, did I forget to take my cynicism-suppressors yesterday.

[M. J. Young]

It seems to me that most people responding to this thread are missing the point of the thread.

Either that, or I am.

(I have not at this moment finished reading it; I will finish reading before I hit post, but I might not fully integrate everything I read into this.)

Ah, I see that Ron has spotted the same confusion. I hope all this stuff I've written is still relevant.

RoboPlayer 2000 is not postulated primarily as an idea for whether the technology might do this. He is postulated as a foil for discussing what it is about players that makes them different from this. The point of the thread, I thought, was to discuss the player.

Fang was on the right track for this when he was talking about reactive versus proactive play. In fact,

Nothing in RoboPlayer's system above bars the program from being one that gives proactive actions high scores.

he was in turn missing the point. Fang's point (I believe) was that the player will surprise you. He will come up with an action that takes the game in an unanticpated direction. Players will do this to you all the time. I've done it to every referee whose ever had me in a game--most recently, I recall, my son was running several of us in what was set up initially to look like an orc stomp. I wondered why orcs who had lived quietly in the mountains for generations suddenly decided it was necessary to attack the elf villages, so when we captured one alive I insisted it be questioned. Sure enough, that revealed that there was a villain behind the orcs, forcing them to attack the elves against their wishes and better judgments. Immediately we bypassed the entire orc stomp and went for the villain. I've had players do this to me, too, pulling the rug out from under me by doing something completely and totally unanticipated.

Roboplayer will never do that to you. You will have provided it with all the options; it cannot invent options which you do not expect. Hey, this is what I hate about CRPGs: I always want to do something that's outside the parameters of play but seems obvious to me. Roboplayer always stays within the parameters of play. He'd be great at CRPGs because his options and the games options are both limited. He would be terrible at RPGs because he couldn't invent an answer that hadn't been considered already.

Now, certainly I've learned how to manipulate players. Every referee has done so. I know the hooks at which they bite, the traps into which they blunder, the scents which draw them. But I think that even the most malleable of players will at times surprise you. That's the problem with RoboPlayer. You know the hooks, traps, and scents, and can infallibly guide it where you want it to go. I've never had a player I could infallibly lead by the nose all the time.

And really, if you did, why would you need him? You could just play his character yourself, and he could go home.

--M. J. Young

[C. Edwards]

This discussion brings to mind the Star Trek: The Next Generation episode where Data and Jordy go to the holodeck so Data can play at being Sherlock Holmes.  Jordy mistakenly asks the holodeck to create an opponent capable of defeating Data, instead of an opponent capable of defeating Holmes.  The result is a super-powered Moriarty with all of Data's knowledge and processing power.

Let RoboPlayer collect enough play experience and I imagine it would run circles around most human players.  No human has the data storage and recall capacity of a computer.  I imagine RP would be ruthless when it came to covering all potentialities; checking for traps, taking prisoners, questioning those prisoners, etc.

The RoboPlayer's downfall would be, as M.J. seems to suggest, its inability to decipher the subtleties of human behavior.  The thing is though, once RP was duped, bluffed, or otherwise led astray, it would never forget and would always attempt to plan for those possibilities.  I would be inclined, personally, to always give RoboPlayer the 0-level Scribe character sheet, just to keep the damn thing from getting the jump on me.

Two things the RoboPlayer really has going for it:

1. No ingrained prejudice about what type of game it likes.  RP is up for anything.
2. Never eats the last piece of pizza.

-Chris

[weeble]

Hello,

May I suggest that railroading is only an issue when those subject to it realise that they are? I do not believe Roboplayer has enough understanding of its world to come to this conclusion and noticeably change its behaviour, but even if it did, such a realisation could be deterministically avoided by the GM who can see into Roboplayer's soul.

Do also bear in mind that the world upon which Roboplayer bases its actions may well be modified by other less predictable players - none of them play in isolation. While the act of creating such a construct may require amazing feats of intellect and technology, the act of outthinking one (or more) which react to (some) stimuli out of your control is an order of magnitude worse.

Regards,

Weeble.

[hyphz]

What I hear is "Using...input, using...information...generated or supplied to...generate rules....  When time comes...look...judge...and emit...."  Everything 'used' must be 'received' first, then acted upon.  That's strictly responding to stimuli, reactive behaviour.

And I think that's all you really can do in non-authorial games.  No matter how original your idea is, it must have a foundation in SOMETHING the GM has specified.  

Mind you, I'm somewhat of the opinion that if you can announce you're doing something that's original and then the GM improvs to fit with what you suggested, you are in fact playing author stance even if it dare not speak its name, as you are authoring the statement "The world allows me to do (this original thing)" into the world.  In fact this seems like a lazy author stance because you author the one statement that gives your character an advantage into the world and then expect the GM to justify it.

I am surprised by your comment about non-authorial games; I regularly launch into all kinds of mischief that were in no way offered by the gamemaster without once taking Author Stance.  That's what I call 'taking the initiative.'  "Can be reasonably entailed" doesn't sound like something a Roboplayer would pursue.  (Likelihood of unmentioned data being outside of acceptable parameters = high; do not explore.)

This seems odd.  When you get up to that mischief, is it in fact unacceptable in terms of playing your character wrong?  Unacceptable in terms of breaking the social contract?  Unacceptable in terms of the premise and focus of the game?  I presume not, since you do not strike me as the kind of player who would do something that truly was unacceptable.  But if it is not unacceptable, then RoboPlayer could do it too, and his program should tell him that it isn't unacceptable.  If RoboPlayer's program makes him think it's unacceptable, his program is wrong.

[clehrich]

RoboPlayer does have certain advantages in a very setting-heavy game, if the setting is very detailed.  That is, if the provided setting is hundreds of detailed pages, then RoboPlayer is actually going to know all of it.  Furthermore, it won't make assumptions based on previous games: if you design a game set in a universe that is cosmetically similar to, say, AD&D, RoboPlayer will not assume that slaughtering orcs is necessarily acceptable behavior.

For me, this shifts the focus away from railroading per se.  You can railroad anyone, although RoboPlayer is unusual in not minding this, because the more you railroad, the more effective RoboPlayer will get.  That is, the more you have pre-determined the possible range of outcomes, the more effective will be a program that similarly limits the range.  But even without railroading, if the focus of play is on interacting with known setting detail, RoboPlayer will shine in some respects.

At the same time, my experience is that the most heavily railroaded games, if the GM is good at it, just like the best setting-heavy games, are the ones in which the players can use the same material against the GM, twisting it in ways not predicted.  Here RoboPlayer would be at a big disadvantage, unless it's just that the GM hasn't though through his own material.  In such a game, however, RoboPlayer might be a useful facilitator.  

This is rather like Anya in the run-up to the destruction of Glory, in a previous season of Buffy the Vampire Slayer.  She complains that nobody is doing anything, and eventually they ask her, "So, what would you do, if you're so smart?"  She produces a whole range of possibilities, all keyed off of prior events in the season: use this or that magic item, etc.  Unlike the regular characters, she has remembered all the provided detail.  At the same time, actually putting together a plan out of this takes other characters than Anya.

[Le Joueur]

Hey Hyphz,

Sounds like we're goin' in circles.

What I hear is "Using...input, using...information...generated or supplied to...generate rules....  When time comes...look...judge...and emit...."  Everything 'used' must be 'received' first, then acted upon.  That's strictly responding to stimuli, reactive behaviour.

And I think that's all you really can do in non-authorial games.  No matter how original your idea is; it must have a foundation in SOMETHING the GM has specified.

I believe you are either creating a false dichotomy or a straw man here.  We can either get into the 'tit for tat' kind of discussion where I keep coming up with more and more extremes ('Okay, how about my paladin sleeps with every woman in the village to forsake his vow of chastity?') or we can argue about what "non-authorial games" are ('I go find the Sherrif.' 'You just created him.').  You can just keep upping the ante ('Every village has women in it and you paladin is male; what isn't founded in GM specification?' or 'you made that up; it's authorial gaming.')

Both miss the point.

We're talking about what people do that Roboplayer doesn't.  If you keep changing Roboplayer's abilities, then you're the only one talking.  I focused on a couple of things you posted:

[*]Using GM's descriptions and game rules as input....

[*]Using some information about a 'character' to play...generate...judgment rules that return [a] score...for potential action;

[*]When time comes to act in the game, look at all possible/appropriate actions in the database, judge each action using the rules in the program, and emit the one that produces the highest score.[/list:u]

This says, quite clearly 'take the safe route' in play.  And that brings us back to the major question you asked:

[*]What can a human player do that RoboPlayer can't?[/list:u]

That's what M. J. was talking about: surprise.  Since we're talking about things like "premise, social contract, [and] group focus," the 'best' answer would to be 'don't make waves,' "emit" no surprises, don't derail the game.  What about M. J.'s 'captured Orc' example?  Since it seemed like it was going to be an "Orc Stomp," following his lead would be less appropriate compared to all the other alternatives.  You clearly specified that Roboplayer only goes with the highest scorer; steering away from the "Orc Stomp" would lower the score, short circuiting the predicted adventure would lower the score, heck 'taking over the group' and leading them somewhere they haven't shown interest should lower the score.  It may not be the lowest scorer, but certainly you can see this kind of 'surprise behaviour' is outside of your posted list of Roboplayer performance parameters.

Any kind of player creativity will carry these risks; Roboplayer has been defined as not being a risk-taker.  This is one answer to your question; "a human player" can be creative.

Mind you, I'm somewhat of the opinion that if you can announce you're doing something that's original and then the GM [improvises] to fit with what you suggested, you are in fact playing author stance even if it dare not speak its name, as you are authoring the statement "The world allows me to do (this original thing)" into the world.  In fact this seems like a lazy author stance because you author the one statement that gives your character an advantage into the world and then expect the GM to justify it.

Let's not get mucked up in another Stance war.

Briefly (and over-simplistically):

[*]Actor Stance = using only character knowledge and character ability to follow character motivation
[*]Author Stance = going beyond that to affect your character for 'outside of character' motivations
[*]Director Stance = going beyond even that to affect the whole game for 'outside of character' motivations[/list:u][/list:u]The way you're using it, the character simply waking up in the morning is Authorial, because you did it not the character.  If I suddenly decide my character has been smitten with the queen and will attempt to seduce her, this is character knowledge and character ability and character motivation and if we're just passing through the capitol on our way to where the gamemaster has sent us, it is most certainly a surprise while being "non-authorial."

Yes, you can argue that the gamemaster has implied there is a queen, by the presence of a capitol and yes, the gamemaster can improvise like a bandit to react to it, but you must agree it runs the risk of breaking the social contract, premise, and maybe even the group focus, all of which make the action rate pretty low in Roboplayer's "'program' of judgment rules" so he won't do that.

That's what "a human player" will do that Roboplayer will not.  Surprise!

Now if you want to redefine what Roboplayer does or does not do differently than the above then there's no way anyone but you can know the difference, why did you even ask?

I am surprised by your comment about non-authorial games; I regularly launch into all kinds of mischief that were in no way offered by the gamemaster without once taking Author Stance.  That's what I call 'taking the initiative.'  "Can be reasonably entailed" doesn't sound like something a Roboplayer would pursue.  (Likelihood of unmentioned data being outside of acceptable parameters = high; do not explore.)

This seems odd.  When you get up to that mischief, is it in fact unacceptable in terms of playing your character wrong?

There is no 'wrong way' to play my character; he's mine!  Is that the real culprit for your original question?  Do you believe that players should only be allowed to play their characters 'the right way' according to the gamemaster?  If that's the case, then there is absolutely no difference between Roboplayer and a human player, expect the human player is capable of making mistakes.  (And this question becomes pointless, so I'm assuming that isn't what you're thinking.)

Unacceptable in terms of breaking the social contract?  Unacceptable in terms of the premise and focus of the game?  I presume not, since you do not strike me as the kind of player who would do something that truly was unacceptable.

I scent the unmistakable odor of a false dichotomy here.  "Acceptable" and "unacceptable" are not clear-cut black and white.  It seems like you've taken us away from the 'what are humans like?' into 'what is free will?' type of question.  By your definition, Roboplayer will never do anything to jeopardize the above (and be boring because of that), whereas a human player doesn't play for 'optimized emissions.'  They make mistakes, they 'emit' surprising responses, they don't do what could be expected; that's what makes them human.

And you obviously don't know me as a player very well.  I don't feel I've really played until I've challenged the gamemaster with something unexpected (but often rational).  Socially, I'm the court-jester; if I don't breach the social contract once in awhile people have to focus on actual social gaffs (I provide them as entertainment which gives everyone an out - laughter).  And if I can't 'take over' the "premise and focus of the game" from time to time, I get bored because I realize I have no real control and am probably being railroaded (gotta test those limits).

Now, is any of that "unacceptable?"  At times, but this dichotomy is pointless; Roboplayer doesn't play 'on the edge,' he optimizes, he goes by 'the safe route,' he does 'what is right' (or you haven't presented him well).  What's important is I'm not a 'safe' player; that doesn't mean I'm totally unacceptable.  That's why it's a false dichotomy; it doesn't cleave like black and white.  I mean, are you asking if players should even be allowed to play 'unacceptably?'

But if it is not unacceptable, then RoboPlayer could do it too, and his program should tell him that it isn't unacceptable.  If RoboPlayer's program makes him think it's unacceptable, his program is wrong.

Hang on a moment there, if you keep changing what Roboplayer is capable of how can we answer:

What can a human player do that RoboPlayer can't?

You keep making the difference less and less.  The only real message I've gotten so far is the gamemaster has a right to say the 'right way' and 'wrong way' to play my character.  Perhaps you could restate the question instead of continually modifying the parameters; I'm getting lost in all these Roboplayer upgrades.

What was it you wanted?

Back to your original questions:

[*]Is it possible for a GM *not* to railroad RoboPlayer?

[*]Is it possible for a GM *not* to railroad RoboPlayer if the GM knows RoboPlayer's program and the program contains no ambiguity? How much of the program can the GM know before this problem arises?[/list:u]

Actually, since Roboplayer seems to be continually trying to 'do the best thing,' I should say that it is impossible to railroad Roboplayer, appealing to the gamemaster seems to be all he's there for.   What would be more appealing to a gamemaster who is bent on railroading than an acquiescent player?  But it can't be railroading unless it breaches the social contract (going against the wishes of the players - if they knew - otherwise it's just Illusionism or Participationism), that can't happen with Roboplayer.  (When comes the point when Roboplayer says, 'hey, I didn't want to do that?')

I can't say this conversation seems to have much point.  I say, 'human players surprise.'  You change Roboplayer and say 'he can too.'  I say, 'but you said this,' and you change it and invoke a false-dichotomy.  Now, I'm saying, 'what are you asking?'  Which brings us back to where we started.

Otherwise, you seem to be asking us to confirm that Author or Director Stance goes against the gamemaster (what no teamwork?); because that's the only thing you've said Roboplayer can't do.  It really winds up sounding like you're asking us to confirm your beliefs that the only thing a human can do is play wrong according to your beliefs.  That's going to stay your opinion no matter what, why seek reinforcement?

Fang Langford

[Mike Holmes]

Hyphz, please stop stringing people along. You confirm to several people that RoboPlayer is allegorical for a certain kind of human, yet you continue to debate Fang on the subject of non-human players.

That makes this at least two threads, if not one that's a troll.

Mike

[Jeph]

Whoa. This would be the ultimate machine in game theory computations. Anyway.

To me, RoboPlayer seems like the ideal GM aid for NPCs: generate tactics and actions, but leave the actual role-playing up to the GM. It can also seriously help a player with strategy. If they watch RoboPlayer enough, they'll soon figure out how to play their PC as effectively as possible. And, conversely, the GM will learn how to play their NPCs that way. But, in games like TROS, do we really want that? In games like Riddle, learning how to not die is part of the core game experience.

[hyphz]

Ok, it seems that the whole RoboPlayer metaphor has got tangled up and confused people, so I'll come out with it directly.  I am, in fact, asking about free will versus determinism.

Just to make sure of this, the real-world determinism argument isn't that we have no choice about what to do or that everything is worked out in advance.  The determinism argument is that we do indeed have choices, but we resolve them with regard to circumstances whose values are all fully bound at the moment when the decision is made.  So, a free will advocate says "He didn't do his essay, because he was too lazy."  A determinist says "He *couldn't* do his essay, because he was too lazy", because at the moment he decided to not do the essay, his laziness was established, as was his decision-making strategy.

In the real world this is an infinite debate, but in RPGs we have a few differing factors that change things around.  First of all, all of the "circumstances" involved are a far more limited range and come from a much smaller range of sources.  Second, as a player you can make choices that you couldn't in the real world, like deciding that you've fallen in love with the queen when previously you only knew there was a head of state of unknown gender.  Thirdly, at least some RPGs which have personality modelling mechanics involve the player exposing at least part of their decision-making strategy for the character at chargen.

My questions then, were:  can a conclusive proof of determinism/free will exist in RPGs?  (I suspect not)   Can a player tactically choose to play deterministically, and is it a good idea for them to do so?  If a player is playing deterministically, is every adventure a railroad?  How many (and what kind of) personality modelling mechanics can a game have before a player cannot avoid playing deterministically if they are to remain within the game system?  

I apologise if people don't think that these are useful questions - I acknowledge that they're fairly disjoined from anything in the real world (thus the theory post).  But I assure you that I did not intend to troll (yea, I know that doesn't mean much because trolls say it too, but I hope it means something)

[M. J. Young]

Part of the problem with this question is that I don't think you can escape the real world level of the matter. That is, with a real player running an imaginary character, you can never get the one hundred percent separation that such an exploration seems to require.

I don't often lead my players where I want them to go; but I do it sometimes. Sometimes it is done by baiting the character; but it is more often and more effectively done by baiting the player. I've known a score of guy players who cannot resist rescuing the fair maiden in distress, even when they're playing female characters. I remember commenting to a friend once that I was thinking I wanted to verse out a certain player's character because I wanted him to test play another world for me, and our mutual friend immediately suggested that the player was a sucker for a trap--it didn't matter what character he was playing, as he would walk right into it with his eyes closed. Sure enough, I set up a rather obvious trap at the next session, and he blew himself up.

I'm of the school of thought that holds that your choices are to a large degree determined by factors of who you are and what you've done in the past; that if you come to the choice between A and B and you choose A, we could rewind time a billion times and run it again and (as long as you were unaware of the repetition and no factors changed) you would never choose B. I don't think that's contrary to free will. In the end, you would argue that you understood the risks and consequences on each side, and you applied your values to the choice, and made your decision. Just because doing that would always yield the same result does not mean the choice wasn't made freely. No one made you choose something other than what you wanted. It's just that what you wanted was ultimately predictable.

Therein lies the flaw in the RoboPlayer example. RoboPlayer is not merely ultimately predictable, he is ultimately predictable by the referee. My players may all be entirely predictable for some divine omniscience who knows them more completely than they know themselves, but for me there will always be that aspect of unpredictability that surprises me. It will come from unexpected sources. For example, I must assume that if I kept running those players into lose/lose situations when they go to rescue the fair maiden who turns out to have been the villain all along, eventually that hook isn't going to work. If I set traps everywhere, the player who always falls for them will become at least occasionally a bit more circumspect. I don't know them well enough to know when they are going to break the pattern and do that which to me is unexpected. It may be completely predictable to someone who knows them well enough, but that omniscient person is not me.

So if you keep leading the players around by knowing them well enough, they're going to realize eventually that you're manipulating them, and this becomes a factor to be accounted when you attempt to manipulate them--knowing at what point to reverse it so that when they see the path you would normally be recommending and they avoid it, they'll be on the path you really wanted them to take. I'm not that good, not even with my own kids. Are you?

--M. J. Young

[Mike Holmes]

Well said, MJ. I'm in agreement on all points.

As I once wrote in an essay on the subject of Free Will, who cares if the universe is deterministic or not? It doesn't seem that way, and so reacting to it as though it were is senseless. You can't determine a course of action based on a phenomenon which is unobservable. Heisneberg is our savior from the enuii of determinism.

Here's an easy test. If you have roboplayer set up, and you give him two routs which are both equally viable choices (which you can determine because you have infinite nowledge of him), which way will he go? How can you predict it? So we see that even with infinite knowledge that the results cannot be determined.

Much less with a real GM who, not being omniscient, has less than perfect knowledge of Roboplayer.

Sure there may be a point at which knowledge is sufficient to make us feel that way. Ignore that feeling and carry on with having fun. Or get less predictable players.

Mike