It’s an intriguing idea, but how would a player communicate their intent to the game, particularly without revealing spoilers or making the interface too complex?
well, you click the peg leg, and you have a popup menu with only: “look at”, “use to…”. If you click “use to”, you are presented a list of objectives: have kate jailed, have the carpenter leave the shop, and a dozen others (probably more in monkey2). . You need to click one. Do you see problems? I don’t see the complexity. About spoilers… if the objective itself is a spoiler, we have a problem. That’s why I wanted to analyze those puzzles where part of the puzzle is to understand what your objective is . Here’s one: use “wanted-poster-with-guybrush-face” to “have-kate-jailed”. (No need to click the leaflet with kate’s face.) Now, Monkey2 does not tell you explicitely that you need to have kate jailed. But my doubt is: maybe it should have? maybe the puzzle is too obscure just because Monkey2 expects you to understand that? How many people solved that puzzle without brute-forcing?
The problem is, that Monkey Island 2 is designed to work with the verb interface. You are now trying to convert the puzzles to your interface - and that will not work (at least not with every puzzle). You will run into such problems like I cited above. And with your approach, the interface will spoil the solutions:
A dozen? I don’t think that this would be funny. 
Beside that, with the option “have the carpenter leave the shop” you give the player the solution to the puzzle. You hide that solution under several others, but you give him the solution. It would be the same if I ask you to calculate 2 + 2 and then telling you, that the right answer is: 1, 2, 3 or 4.
Your Interface might work in a new game, but I doubt that it work with Monkey Island 2.
You might be right. I was assuming that the core of the puzzle is to understand that “you need to break something made of wood in town”. I.e. that the core of the puzzle was to understand how to make him leave. But if the core of the puzzle is to understand that you have to make him leave, the puzzle loses something important, implemented like this. It is a matter of understanding first what the puzzle actually is.
Edit: if we want, we can call the objective “pick up the nails and hammer” instead of “make the carpenter leave the shop”. So it’s not a spoiler anymore. This means you are required to do “use peg-leg to pick-up-the-nails”. But it would be too far fetched, right? It would require too long a chain of reasoning. But then, why is it not too far-fetched in monkey2? That’s exactly what monkey 2 asks you to figure out. I suspect if the puzzle is both to understand what to do and how to do it, if both things are mysterious, we might have a problem in the puzzle in the first place. (how many people would solve it in the “proper” way?)
I wonder why you feel so. Why is it funny if you have 30 items in your inventory, but not funny if you have a dozen (or 6, or 30) objectives in your “objectives inventory”? I don’t see a difference.
Ah, you would replace the inventory with a “objective inventory”? I thought that you would like to make this context sensitive? In the latter case: With the verb interface, the player is able to scroll through the 30 objects and think calmly about them. In your version, the player clicks on an object in the scene. He then is suddenly overwhelmed with 30 possible objectives. Not only he has to read all these objectives, on the next object he gets 30 new objectives. So in the sum we have a large amount of objectives he has to deal with.
I see. So the problem is that I presented this to you as a popup list that only appears after you click the room object. This is not necessary, it’s an implementation detail. So if we do like monkey 2 (objective inventory always on screen), it would be ok for you?
Yes, that could work. To give a more specific answer, I would need to play a prototype.
@seguso: you could do it using a “thought space” interface similar to the one used in “Sherlock Holmes: Crimes and Punishments”.
In the scene in which you realize that you need a hammer, you can add the thought “I need a hammer” to the “thought space”.
In the scene in which you see the hammer but Woody doesn’t let you to take it, you can replace the thought “I need a hammer” with “I want Woody’s hammer but he doesn’t let me take it”.
If the game is well designed, it gives it to you an important information that is needed to make a deduction. For example, if there is a sign “Door-to-door repair services”, you can add the thought “Woody goes out to do emergency repair services” to the “thought space”.
If the player makes the deduction, he can combine the two thoughts “I want Woody’s hammer but he doesn’t let me take it” and “Woody goes out to do emergency repair services”. This automatically creates the thought “I have to lure Woody out of his shop”, that replaces the two thoughts that led to its creation.
And so on…
Basically, the thought space contains both objectives and information and it’s up to the player to combine them. Thoughts can be automatically added by the game or by the player (for example, if you need an object but you can’t take it, you can click on the object and select the voice “I want this”) depending on the context.
1 Like
Thank you. This is “forward reasoning” (as opposed to “backwards” reasoning, starting from the goal). With the purpose of allowing the user to generate the correct objective, instead of just giving it to him. (So he will be able to generate the objective "make woody leave the shop " from the objective “take the hammer”.) Right?
I had thought of something similar. (also in Blackwell 1 you can combine concepts to produce inferences). However, I am reluctant, and I’ll try to explain why. Before introducing this complexity, I’d like to be sure 1) the problem is not in the puzzle in the first place (see my response above to Someone-- I suspect a puzzle where both what you need to do and how you need to do it are mysterious might be too obscure). 2) even if the puzzle is not flawed, I need to be sure the current system is not sufficient to represent the puzzle. With the current system, I could still give as objective “take hammer” and require the user to compose “use peg-leg to take-hammer”. Why is this not ok? Maybe because to connect the peg-leg and the hammer requires a chain of reasoning that is too long? But then, why does the sherlock system help? It does not make the chain of reasoning shorter. Maybe it is supposed to be an aid to reasoning? But the player is supposed to do the reasoning in his mind. What’s the purpose of first making a chain of reasoning that’s too long, and then giving him a tool to help him manage chains that are too long?
Right, but things change depending on what you mean by “generate”.
If you mean “giving to the player a way to define a goal by scratch”, for example writing it, then this solution might be quite difficult to implement.
Instead, if you mean “letting the player to combine the two correct thoughts and assume that he’s combining them because he has made the deduction”, then the goal “make woody leave the shop” might be automatically added by the game after the player has made the connection.
Are you referring to the specific puzzle of the hammer? If so, why do you want to design a system able to represent and handle this specific puzzle? Every system and interface is able to handle only some kind of puzzles, you just have to understand which kind of puzzles you can handle with the system that you have in mind.
I would say so, yes.
That system is irrelevant to the kind of solution that we are discussing. We are discussing a system that aids the user to make deductions to traverse the puzzle dependency chart, so it makes sense to visit each node of the chart, avoiding too long jumps.
The Sherlock system is a completely different thing and I cited it only to provide a conceptual/visual image of how a “thought space” might be.
Well, I’d like my system to be able to handle any puzzle that is desirable to have. (this excludes those puzzles which require a chain of reasoning that’s so long that nobody could realistically do that reasoning). Since I found a puzzle in monkey2 that seems problematic, I need to understand if the problem is in my system or in the puzzle. That’s why I am focusing on that puzzle. As a test. For now, I believe my system highlighted a flaw in the puzzle. [Edit: no, because the puzzle is actually two puzzles, each of which is not too long]
So this is the context… Now, you suggested a “system that aids the user to make deductions to traverse the puzzle dependency chart”. (I called this system “sherlock system” and this led to misunderstanding.) Now, I have tried to raise doubts whether this system is desirable. To express my doubt, the best thing is to reason by cases: either it is too much to ask the user to connect A with B, or it is not. Now, if it’s too much to ask, then the problem is in the puzzle. If it is not too much, then why give him a tool that allows him to generate an intermediate goal “C”, and tell him “good job man, the reasoning so far is correct”? This way we are making things too easy for him. Because we are under the assumption it is NOT too much to ask him to connect A and B unaided.
(If this seems too abstract, replace A with “peg leg”, B with “take-hammer”, and C with “make-him-leave-the-shop”. )
Am I making sense?
I believe you are overthinking this problem and that your own stated assumptions are false. For example:
either it is too much to ask the user to connect A with B, or it is not.
Or, perhaps it is obvious to some, not obvious to others, and perhaps just a tad strained logic to some others. Humans are unpredictable.
The focus shouldn’t be in forcing (there’s that word again) users into thinking in a specific linear way but guiding them into reasoning the solution and advancing the plot (if indeed the puzzle is a plot device and not just a “gate”). How they arrive to it, should not matter.
I submit that if the particular thought process or order of logical connections required to complete a puzzle is germane to the plot, then you have put yourself and the player in a precarious position that may lead to frustration from both sides.
I will be honest, sometimes when I play adventure games (even Thimbleweed Park), I may encounter a puzzle or solution which may not be absolutely clear to me outside a greater context, other than to know that it “means something.”
It may be later on that I encounter a situation which triggers the recognition, in a sense “clicking” all the previous pieces together. At this moment I will go “aha!” and instantly “get” the previous joke or plot element. Sometimes I’m just slow to process some things.
Is this optimal? Perhaps not. Does it really matter? I don’t think so; it did not diminish my overall enjoyment of the game.
More to the point, who are you (or any designer) to claim that my (eventual) mental deduction and thought process is wrong?
Now, if what you are saying is that there is a puzzle which most people can’t solve because the mental gymnastics required to figure it out are too esoteric, weird, or just plain illogical (yeah, Sierra, I’m looking squarely at you!), then the problem is not the player’s powers of deduction or processes thereof – the problem is in the puzzle.
Sometimes we have to recognize that some things that make perfect sense to us may not be as sensible as we think. Shared personal and cultural experiences have a lot to do in making those connections obvious amongst different people.
A lot of jokes and puzzles in Day Of The Tentacle, for instance, are based on what I call cartoon logic: absolutely silly and absurd situations or gags which have traditionally been depicted in old Warner Bros. cartoons and their ilk, as a force of comedy.
Given that a large part of the American population had been exposed to these gags via television or film, then it’s a shared experience which the game designers can leverage. Put the game outside the context of the 1980s American TV audience and all you have is a bunch of illogical and completely random antics.
dZ.
I don’t know what you mean by “puzzle”, to me it helps to see them only as deductions or intuitions.
If the player has made the deduction, he just has to combine three things: “peg leg”, “saw” and either “make Woody leave the shop” or "Woody goes out to do emergency repair services”.
Interesting. In other words you are saying I am thinking this the wrong way, because it’s not possible to really isolate “puzzles”, or define what a puzzle is. because it might be all a single big “puzzle chain”, where each node is at the same time an end and a means to achieve another end?
Edit: sorry, I think I misunderstood! You are only saying that the interface you suggest allows you to draw forward inferences “if X and Y then maybe Z”. Like “I need to take the hammer + he usually goes away to repair stuff ==> maybe I need to make him leave the shop”. I get this. It makes sense and is interesting. But then my reply above is relevant. If you allow him to generate these inferences, you are implicitely telling him he is on the right track.
I think we can’t escape the fact that we first need to decide what the puzzle is:.
We have this puzzle chain:
must take hammer <---- carpenter must leave <----- must saw peg-leg
Now, how many puzzles are these? are they two separate puzzles, or is it one single puzzle that connects hammer with peg-leg? If you decide they are two puzzles, then forward reasoning is ok. But if it you decide this is one single big puzzle, then, by allowing to generate the intermediate, you are only giving the player a big hint.
it would help me to understand the sequence of clicks needed to generate the clickable object “make woody leave the shop”.
…then you decide to treat them as two puzzles, right? 
What’s the dilemma, here? Are you asking which one of the two scenarios works better in most cases? It’s the first one. You solve one “<-----” at a time.
Forget about it. I have just realized that you need only the other one, which is created just by observing the environment: "Woody goes out to do emergency repair services”.
The player connects “peg leg”, “saw” and "Woody goes out to do emergency repair services” and this connection is enough to assume that the player made the correct deduction.
In other words, the goal “make Woody leave the shop” is in the player’s mind but not explicitly cited in the interface.
1 Like
Right! We just need to call them two separate puzzles, and there’s no problem! So your forward inference suggestion works very elegantly!
Sorry for being so verbose, but the issue is confusing.
Now, the only thing that I don’t follow is the implementation detail: it seems you want to implement this by combining three things at once: click “peg leg”, click “saw”, click “woody-usually-goes-out”. And then click “combine all”. Right? (so, exactly like monkey 2 but you are also requiring a third object.) (clicking the saw still seems a bit superfluous to me but it does not matter).
It seems more natural to me to do it like this, with two separate inferences: first combine “must take hammer” AND “woody-usually-goes-out”, and it generates the inference “must-make-woody-leave”. Then combine “must-make-woody-leave” AND “peg-leg”, and guybrush does the action.
It’s not completely clear to me which implementation is better. Anyway, I’m convinced that the “forward reasoning” approach works well.
Moreover, it just occurred to me that my original approach (clickable objectives and backward reasoning) is equivalent to the forward reasoning as I just described it. (If you reread the above, these are simply objectives that are generated from other objectives.)
My concern here is what if the player doesn’t know that the carpenter can be lured out of his shop? By looking at the sign and the hammer, it sounds like the game would end up telling the player that the carpenter can be lured out of his shop without the player necessarily realizing that first. Furthermore, what if the player has no idea how to lure the carpenter out of the shop, and just happens to get lucky by combining “must-make-woody-leave” and “peg-leg”? The player may very well end up completely surprised when Guybrush decides to saw off the peg leg. For instance, what if they thought they could vandalize the peg leg by carving on it with the knife? What if they thought the peg leg could be simply swapped with the broken oar, perhaps with the expectation that it would also provide them with a replacement item to gain access top the tree?
At least having to use the saw on the peg leg means the player has a reasonable expectation that they’d be using it to cut the peg leg.
Thanks for raising so many good points, Caven.
I think we need to backtrack a bit and make it clear what problem we’re trying to solve. The problem is that most users will combine saw and peg leg just because one can cut the other, without knowing why this is useful. We must think in terms of probability: the probability that this happens is high because these two objects are “compatible” with each other, so it is very likely you will combine them without knowing what will happen. Instead, the probability that you will combine “make Woody leave” and “peg leg” is much smaller, because at first sight these objects have absolutely nothing to do with each other.
You also raise a different problem: in my implementation , and in lowlevels implementation, there is a clickable fact: “Woody usually goes out” . What if understanding this is part of the puzzle? If the game automatically creates this clickable fact, it’s a giveaway.
I think this problem comes from the fact that we are constantly shifting what the puzzle actually IS, and how many puzzles there are: whether they are several puzzles , or it is only one big puzzle. we can’t have it both ways: if we think it’s one big puzzle, it’s a giveaway to even generate “must make Woody leave”, let alone “Woody goes out”. So it seems to me you are assuming it is one big puzzle. But if we assume that, then it’s acceptable to implement simply like this: use “Peg leg” to “Take hammer”. Hiding all the intermediates. Why?. Because now we are under the assumption it’s not too much to ask the user to connect these two things directly.
As alternative, If you think these are several puzzles, not just one, we could give the user a way to generate the fact “Woody usually goes out for calls” , instead of just giving it to him. This might be possible , but I would not focus on that, because it does not seem to me this is an important part of the puzzle.
I might be overlooking something. I will think about this more. Since I’m finally going for vacation, I’ll have time 
In my opinion you should stop thinking about what theoretically is a puzzle and start building a proof-of-concept of the mechanism that has been discussed.
The user has a list of already observed/taken objects and a list of thinks that he knows, that the system creates only when the player observes something (or interacts with the environment in some way).
The challenge consists in 1) telling the game which action the character has to do and 2) providing a piece of information that is connected to the action and that (in the mind of the player) justifies it. It’s a challenge because it requires a deduction.
In the specific case, the objects are “saw” and “peg leg”. The piece of information is “Woody goes out to repair stuff”, which the system creates because the player has observed the object “Sign”.
If the player understands that he needs to lure Woody out of his shop and connects the three things (“saw”, “peg leg” and “Woody goes out to repair stuff”) then he has solved the puzzle.
The puzzle can be difficult because the player has several objects and several pieces of information to choose from and he has to connect the right ones.
For example, when Guybrush tries to take the crypt key in Stan’s shop, the system creates the information “Stan doesn’t let me to take the key”. When the player realizes that he has to close Stan in the coffin, he connects the three items: “nails”, “coffin” and “Stan doesn’t let me to take the key”.
1 Like