Hi everyone,

(1)

as I mentioned earlier I would like to give it a try to make MoM a halfway decent solo-game. There have been a couple of various attempts being discussed and available on the BGG-forums. With one exception there was no solution that really convinced me. Most attempts were about AI-ing the Keeper; and the best effort to AI the Investigator was a self-made story.

(2)

What I would like to achieve, is simply this:

a) I want to create a set of tools which give a solo-gamer the opportunity to play the more vital role within the game (the Keeper) in solo-mode. This means I need to find a way to properly AI the roles of the Investigators. Thereby the goal is crystal clear: In the end it should not matter how many Investigators are on the board.

b) Also I would like to develop a solution which allows you to play the official maps and stories presented in the core-game-set. This aim includes that the rules to AI the Investigators should not have any effect on the general rules of MoM. So basically all this is about playing MoM by the rulebook without changing anything - that is with the exception that the Investigators are not played by different humans.

(3)

Now, some might wonder: Why the *peeep* does Mad want to achieve this? MoM was designed as a multiplayer game. Also there are already some solutions available. Why then?

Well, there are a couple of reasons valid for me:

a) My gaming community is scattered all over my countrysides; and we only randomly and seldomly happen to sit together to play boardgames together.

b) Since I am the one owning MoM, I am the one who most often plays the Keeper.

c) Since we play the game so randomly, I don't get used to it.

d) Conclusion: Giving me a game-mode in which I my anticipate the moves of the investigators, and by which I may "train" the game-mechanics - and all this while playing the intended scenario -, this will help me a lot to not utterly fail in the role as the Keeper, as I did in the past.

e) Finally: Sometimes I happen to write small stories for friends or just for the sake of writing. And since I have a game at hand that delivers more then just rules and miniatures, this game may be a proper source of "telling" stories ...

(4)

Also people may wonder: Why AI the Investigators?

In my opinion the roles of the Investigators are by far easier to AI then the role of the Keeper; each Investigator sticks with her two moves and her action per turn; the only two random things happening are not knowing what is hidden in each room and what will be the result of any dice-roling. In contrary the Keeper has to deal with a lot more variety throughout the game - especially regarding drawing different cards from different stacks for different occasions. Besides this the Investigator has a pretty good position to anticipate the moves of the Investigators because the clues lead the Investigators through the game, and each turn the activities of each Investigator is very limited. Also the Keeper knows the setup of the game, therefore the order of the clues and the end-game-scenario. Considering all this I assume it to be a lot more logical to AI the Investigators, and not the Keeper.

(5)

Now, what is going to happen in this topic?

First of all I will try to give you a step-by-step diary (I would like to call it that way) about my thoughts and ideas and especially the reasoning about any attempts so far. Also I would really appreciate any comments, especially the ones supporting the goal or aim of this project. Thank you so much in advance!

(6)

IMPORTANT NOTE

Since I have to analyze a few things within this game, chances are good that there will be some SPOILERS about the game. Especially my first posting after this one will include things about clues, the various Mansions and the "routes" within those Mansions. If you would not want to learn things, you will have to find ways to ignore such in formation ...

All the best!

Mad

Hi everyone,

so, let's get started. SPOILER-ALERT: This posting provides information about the placement of clues within each Mansion along with the "optimal" route to get there as well as the maximum amount of time available for each Mansion of Madness in terms of game-rounds. The graphics are of "bad" quality insofar that you may not read the content of various cards; also the qualities of the maps are not optimized for printing in order to prevent people from doing stupid things.

(0)

In order to find out, how to approach the AI-ed Invstigator, I need to clarify or measure "time-issues" within each single game. In each Mansion of Madness we have a fixed "time-schedule" (the event-deck) and Investigators being capable of moving each turn three places. The faster they run through each Mansion of Madness, and by that the faster they get to the clues, the sooner the endgame takes place. That's why the first things I wanted to know are these:

What are the fastest as well as longest ways to run through each Mansion; and what schedule was pre-set for each story with the help of the event-deck? At first I will check each Mansion of Madness, afterwards I will try to figure out, if these results are open to any generalization. Important: This does not consider any hindrances placed by or activities of the Keeper against the Investigators; so this here in this posting would be nothing else but probably the optimal "run" ...

Please keep in mind: Although I really tried hard not to confuse things, it might have happened that I twisted something or considered something somewhere not properly. Therefore the results may be not 100% correct; but I reassure you that I put a lot of effort into it; so the results will be "fair enough".

(1) The Lynch Mansion

a) The Map

Within this picture you see the setup of the Lynch-Mansion; on it I have marked the possible places in which the clues are located. I tried to connect the various clues according to the way they are supposed to be collected. In order to limit the confusion I tried to give each single connection between two points a single color-code (exception: the green code - this is simply the same route for both Clue 1 positions). In order to find out, how many moves plus a single Exploration are needed, you only need to follow each line and count each space you pass. You will see: This colorful picture gets less confusing. Try it ... !

So - for example: In order to get from the Start to the first clue (Clue #3) you would need to follow either the red or the yellow line; since each Clue #3 leads to a different location for the Clues #2, I kept the color-code the same. Which means: If you play Clue 3A, you follow the yellow line; therefore an Investigator would need to move 4 spaces and do 1 Exploration to uncover the Clue #3A - this will cost him 2 turns; afterwards he moves on - 8 Moves plus 1 Exploration for the Clues #2 at the end of the yellow line. From here you would follow the green line first - being the same route for any location of the final Clues #1, and after 3 Moves on the green line, you would have to follow any one of the blue lines in order to approach the final Clue #1. The light blue line bases on Clue #2A and leads to the Garden, where Clue #1 is located; the dark blue line is based on Clue #2B and leads into the cellar. (the color-coding varies from map to map slightly.)

(This is how the next maps for the other Mansions work accordingly; since they are more complex, the number of lines is bigger and may (at first) lead to some confusion ... In trust we trust ... I just hope that I did not confuse things too much although having double-checked each map a couple of times).

In Short for the Lynch Map - with each Exploration-Action uncovering the next Clue:

All in all and depending on the location of the Clues you may run this Mansion in 7 or 8 rounds as an Investigator until the final clue is revealed.

This does not consider the endgame. None of the Goal-Cards sets a time-based pressure on the Investigators; but it may be important to know that with the revelation of Clue #1X the Investigators only need to move 7 or less Moves in order to leave the Mansion, which is only a three turn thing.

b) The Event Deck

The Event Deck for the Lynch-Mansion "only" lasts for 16 gaming rounds. Therefore the Investigators would only need about half of the game in order to uncover the story as well as the goal-card of this Mansion of Madness. Again: This is only true, if you move in the most efficient manner through this Mansion without any complication.

(2) Sanctum Mansion

a) The Map

Not let's have a look at the second Map:

As you see there is a lot more "trouble" going on; a special note about this map is the ladder, which appears in the Clue#2 location - changing by depending on where the Clue #2 is placed. This may save the Investigators three vital turns of moving. Considering this, you may have a game-session which may vary a lot depending where the Clues are located.

The shortest game possible would only have 20 Moves + 4 Exploration = 8 rounds, while pretty often you will have a game between 30 Moves + 4 Exploration = 12 rounds and 35 Moves + 4 Exploration = 13 rounds until you have uncovered Clue #1.

(It is pretty amazing to see, how wide this range of game-durations is ...)

If you need to flee the Mansion, the Investigators would need at least three rounds to get away.

b) The Event Deck

The Event Deck for Inner Sanctum ends with the ending of round 20. So, even if you take the longest route through this mansion plus the goal to flee, you still have a "buffer" of about 4 rounds.

(3) Blood-Mansion

a) The Map

The core-game offers one huge "garden-map" - the Blood-Ties Mansion:

As you already see there seems to be a certain route or axis within this map that seems to be the "orienteering" in it. Things swirl around it.

The shortest possible route would only have 11 Moves + 5 Exploration = 6 rounds, while the longest game may take you 18 Moves + 5 Exploration = 8 rounds in order to reveal Clue #1.

In one specific endgame the Investigators need to get to a specific location on the map; this may take them 5-7 Moves in addition to what they moved throughout this Mansion of Madness.

b) The Event Deck

Considering this rather short line of actions of the Investigators, the 19 rounds of the Event Deck are pretty much. I crunched the numbers a couple of times; I so much hope that I did not do anything wrong; edit: but the numbers may vary slightly by one or two Moves.

(4) Classes-Mansion

a) The Map

A map that divides the MoM gamers' community a lot seems to be this one:

Compared with the Sanctum or Blood Map the first impression of this map is: it is not that big, is it? And it is not; but still the Investigators have to move a lot, especially with the Clues in the bottom-right-corner. The fastest game lasts 31 Moves + 4 Exploration = 12 rounds, the longest gets you into round 14 with 38 Moves and 4 Exploration, until you reveal Clue #1.

Depending on the final clue you may need to spend 4 or 5 additional rounds to leave these Mansion of Madness.

b) The Event Deck

The Event Deck covers a time-span of 20 rounds. So with the fastest set-up available you may have a buffer of 4 rounds; else it may get very, very close ...

(5) Green Eyes

a) The Map

The core-game offers you one very huge map that uses most of the tiles available.

And I am very happy that the route may take you through most parts of the building. But time is still an issue - or especially because of that an issue?

If you run the fastest possible route, you still need 47 Moves + 6 Exploration = 18 rounds; in contrary the longest route will take you 50 Moves + 6 Exploration = 19 rounds. The difference is not much in order to receive Clue #1.

Fleeing may take you 5-6 additional rounds.

b) The Event Deck

The schedule for this map is tight: 25 rounds.

(I have not yet played this scenario; but simply considering these numbers, I think this may be a very painful experience for Investigator-gamers.)

(6) Generalization possible?

Plain and simple: No. Looking at the different clue-set-ups on the various maps with eventually major impacts on the amount of time the Investigators have until Event #4 or the goal-card is revealed, it is not easy to draw any general conclusion out of this information - except for one: Each map is set-up in a way that the Investigators could and most often will find Clue #1 before the 4th Event Card is drawn. Each of those (Clue #1 or 4th Event Card) reveals the Goal Card and introduces the end-game. But I need to admit: Especially the fifth map is very tight on this; and I remember playing the Inner Sanctum as well as the Blood Ties with Event #4 being played before Clue #1 was revealed. So theory and the real game-results may vary a lot.

(7) What's next?

Now knowing how "open" or "seemingly unlimited" the time-line of each Mansion may be, I need to get some more ideas about the actions of the Investigators. So basically It looks like I will "force" each AI-ed Investigator to move AND act on each turn with choosing "running" only if nothing else seems to be valid. It is obvious that Exploration - as intended by the game - will be the most highest priority-action of any Investigator as soon as he/ she enters a place that was not explored yet. But what about the other Actions available?

Ok, it's cool. The rule-book is crystal clear on that: Fight, Use an Item-/Spell-/What-ever-Action ... But how about especially the once per game-session actions of each Investigator? And would it be possible to prioritize the different Actions available? I keep you updated.

tl;rd

All the best!

Mad

Hi.

So soon I need to comment, or better: slightly correct my numbers, because I did not consider the breaks between the rounds/ turns thoroughly enough.

The minimum or maximum amount of rounds needed for Inner Sanctum are: 8 rounds to 12 rounds, if nothing hinders or hampers the Investigators.

The minimum or maximum amount of rounds needed for Blood Ties are: 7 rounds to 9 rounds, if nothing hinders or hampers the Investigators.

The minimum or maximum amount of rounds needed for Cursed Classes are: 12 round to 15 rounds, if nothing hinders or hampers the Investigators.

The Minimum or maximum amount or rounds needed for Green Eyed Boy are: 15 rounds to 20 rounds, if nothing hinders or hampers the investigators.

So much for the effort I put into it before I posted these corrections But anyway: The changes are minor and don't change the overall outcome of the result (except maybe for the Green Eyed Boy; and I am so glad that at least in one Mansion I did not get confused with all the numbers ... ) ...

All the best!

Mad

Hello everyone,

it is time to go on. After my insane attempt to "analyze" the speed-optimized routes through the various mansions in order to understand how tight or close the "schedule" in each Mansion of Madness actually is, the next step has to be to understand the investigators' actions during each round or within each game-session. In this posting there will be no spoiler-alert, because I will only refer to things that players already know when they choose and and when they play their investigators. So at first I will give you a very brief overview over the various actions an investigator may take within each turn; afterwards I wish to try to "grab" the rational behind the individual investigator's activities; in the concluding part of this posting I will relate these different actions to the general gameplay of Mansions of Madness. And finally I provide you with a first beta-version of an "AI-ed action-plan" for investigators. But before that let me explain, or better: describe the ideas and thoughts I had to spend some time before I finalized the beta-version-attempt of this AI ... If you don't want to struggle with the wall of words, just jump to the last part of this text.

1. What does an investigator do during his/ her turn?

Basically the rules only allow the investigator to do three things during his/ her turn: Move, move once more, and then take an action.

The action may be only one of these per turn: explore a room, attack a monster, use an item or spell, use a once-per-game-session-special-action of his/ her investigator, hide, move a barricade, or finally move another space.

Maybe you recognize: There is a huge variety in actions you may choose from ...

2. Is there kind of a general idea or rational behind the investigators' actions?

a) Simple answer ...

Plain and simple: Yes. At least, if you understand each different activity as an action in the broadest possible sense, which by the way would be or actually IS the term for it within the game. Action. Regardless of what you do, you either move, move, or you take any kind of action. And that is all. And that does not help us here, does it?

b) ... but!

But there is a huge but. For instance there is a huge difference between just moving a third place and exploring a room; also there is a difference between attacking a monster or using a once-per-game-session-action. Also you need to consider that in some cases you may spend skillpoints in order to support certain activities. And this is actually the tricky part about what I try to do here.

Remember: I wish to achieve a solution by which I give the investigators an "AI". And this AI needs to fulfill a couple of different tasks - it should make the investigators capable of exploring the mansion, successfully passing any challenge I as the Keeper throw at them, and finally using their "abilities" at least in a valid, but not necessarily perfect way.

Now, what I did then was to translate the various general actions and special actions of the investigators and/ or their items. And basically I came up with this - let's call it - code:

A - Explore
B - Move
C - Attack
D - Heal
E - Gain Skillpoints
F - Something else or
G - Passive/ Protective items.

And each of those you may consider as either a general gaming activity an investigator may use each turn again and again; or there are some "things" investigators may only do once per game-session, because each of these is a special ability of the individual investigator. To give you an example:

Jenny Barnes starts the game either with a special letter or a .45 gun, which depends on the "strength"-trait of her character; each of the "intelligence" - traits provide her with a once-per-game-session-action, in her case either a bonus against horror or a bonus for a single attack. According to "my code" I mentioned above this would mean: .45 would have the code C (attack); the letter would be a combined DE (heal sanity and gain skillpoints); the once-per-game-session-action would either be a combined FB (something special plus move) or a C (attack bonus).

And this you can do with any other character as well. Wish more examples? Just yell at me.

Important note: This I only did in order to get a "different" understanding of the actions, which was supposed to help me in finding means to prioritize ingame-actions ...

3. Now, yeah, ok. But what is all this good for?

a) The plot

Before I ride this horse about actions dead, let's take a look about the general game-session. We know that each game-session of Mansion of Madness is basically "structured" like a story. There is an introduction after which the investigators enter the Mansion of Mansion and start their adventure in this mad house. Now chapter by chapter, "titled" by each clue, they follow the storyline, figure the problem and try to pass the overall challenge in the phase of the end-game.

b) Why is this "important"?

We also know that investigators are only successful, when they use their actions - the general game-actions as well as the special item/spell/investigator actions available. In order to make all this work out fair enough (again: not necessarily as perfect as possible), it is pretty useful to get away from the general idea of "action" and prioritize those ones according to what this game is all about.

So here we are: Priority 1 could and should be Explore. And then? Priority 2 should be Move, in order to get on with a decent pace. Attack should be something that is not a top priority action at all - unless the goal card says something else or combat is something an investigator could not evade anyway. So priority 3, 4,5 or 6 for Attack then ...

c) But wait! What, if ...?

Yes, good point. What, if an investigator shares a space with a monster and would need to Explore or Move because of the higher priority instead of attacking? Would Attack become the highest priority action? And what about Evasion? And what ...?

This is a subject I thought about a lot. Naturally there is no sense for investigators with low abilities to dive into combat by any means. And naturally there are some investigators which have "Mortal Combat" written right on top of their faces ... So this is the point where the ability scores of each investigator come in very handy. The final result (beta-version) I will provide in the last section of this topic. At this point I just want all of you to know: This is probably one of the very few things that make this project kind of very complicated; and until now and from now on you may start to realize what different directions one might choose to proceed ...

d) Proceed, OK. Yeah. And what was this stuff with the end-game all about?

Back to the actions again: Since each investigator has his/her once-per-game-session-action, it should be important to consider that these special actions should be used especially during the endgame, when things start to become tough. With only two or three exceptions most of these special actions give each investigator a pretty good advantage either in combat or to re-gain skill-points, which again may be very important in the final battle or struggle of what kind ever ... So here we are - in the (beta-)end:

4. AI-ed action plan

I hope I have not lost you by now. As you see things are not as trivial as one might guess. And if you consider the various solo-game solutions on BGG you will recognize: These ones are anything else but "simple" solution aiming at solo-ing MoM. And so - surprise! - this here cannot be as easy as one might wish. As you have seen by now I tried to get a different or better understanding about the actions; and also I tried to relate it to concepts of a game-sessions as well as ability-scores of the investigators'. And finally I concluded all these various aspects to this "simple" solution:

• General cause of action - Beta-Version:
  
• Priority 1: If an investigator shares a space with at least one monster, check the combat scores of the investigator on his trait card "strength"; if the investigator has an Marksmanship lower then 5, he evades and moves on with the priority 2; else jump to priority 6.
  
• Priority 2.1: Anytime an Investigator enters an unexplored room he uses the action Explore.
• Priority 2.2: After having successfully explored a room an investigator moves on.
  
• Priority 3.: The Investigator uses his spell or item to heal himself or another investigator in the same space; the investigators will move to each other in order to do so.
• Priority 4.: Investigators may use any action (moving barricades, hiding, using special action-abilities, firefight, etc.) which makes most sense in a given situation. With the excpetion of Kate Winthrop's "Theories"-trait or Gloria Goldbergs traits no special trait-action should be used before the start of the end-game unless it is all about the live and death of any investigator before that.
  
• Priority 5.1.: Investigators move into the direction of the next clue.
• Priority 5.2.: When an investigator has moved both his moves, has not yet used his action and has not yet reached an unexplored room, he uses his action with a third move.
• Priority 5.3.: Investigators falling back will follow their comrades as fast as possible.
• Priority 5.4.: If an investigator, who fell back, gets attacked, the other investigators will get back to him and help him.
• Priority 5.5.: Investigators try to share the same spaces in order to exchange/ trade items suiting their skills and traits best.
• Priority 6.1.: Fight the monster.
• Priority 6.2.: If the monster is dead, move on as described in priority 2-5; else: stay in the same space as the monster in order to fight it.

As you see, it starts with combat, it ends with combat, and inbetween there seems to be an approach from most important to least important ... This "concept" is basically kind of a "guideline" (You have seen Pirates of the Caribbean, haven't you?) and needs some play-testing. Also I did not mention skill-points in this posting - since these are sepcial. My next posting will be all about them, adding another level of complexity to this AI.

At the moment I am pretty happy about "this" result. It does not change any game-rule; it sticks with the basic ideas about what investigators are "meant" to do; and since movement aims at following the direct route along the clues, there should be no trouble with the "schedule" within each Mansion of Mansion.

Now ... I wrote so many things ... Any ideas, thoughts, suggestions, criticism you wish to share? Thank you in advance ...

All the best!

Mad

tl;rd

Hi everyone,

this one is going to be short. I will tell you about my plans how to include "skillpoints" into the AI. Since - again - this is all about the investigators there will be no spoiler-alert in this posting.

Important: With the last postings I basically covered the major issues I had while I was thinking about AI-ing the investigators. Therefore I will conclude this posting with good news!

1. What do we know?

We know that the number of skillpoints available to each investigator varies between 2 and 4.

We also know that depending on the chosen traits most investigators are able to "regain" skillpoints during the game - either as an action made available with an item or as an once-per-game-session-special-action.

Besides this the rules tell the players how to use the skillpoints. There are two ways to do so:

a) You spend a skillpoint in order to add your luck-score to any other attribute before you roll your die;

b) You may spend a skillpoint in order to add your luck-score to your intelligence-score while solving a puzzle.

2. What is the issue here?

The issue with AI-ing the skillpoints is very simple: When do I spend skillpoints of an automated investigator. I could spend them as fast as possible. But in the beginning of each Mansion of Madness most tests and challenges have little impact on the game; it really makes only little sense spending them too fast. And although some characters have the "action" to regain skillpoints, it would always cost them an action and therefore probably another game-turn to go on with the story. Therefore I need to figure how to implement the usage of skillpoints for AI-ed investigators.

3. A Beta-Solution

At the moment I consider two things:

a) About using skillpoints on attribute-tests I would choose the characters to use them not earlier then before clue #2 of 3 or 4, or clue #3 of 5 or 6 is revealed/ explored. Also the skillpoints should only be used on those tests with a combined (attribute+luck) chance of success below 80%. That saves the clue-points for the second half and end of the game and makes them available on rolls with low attribute-scores;

b) about using skillpoints on puzzles the usage depends on how likely a character may solve a puzzle by adding his luck-score; if it is possible to finish a puzzle with spending a skillpoint and adding the luck-score to the number of attempts, then the investigator will spend his skillpoint; otherwise he will only spend his intelligence-score on the puzzle and then moves on according the general action plan mentioned in the last posting.

Now, about b) I have kind of another issue here. According to the "best routes" through the mansions (my first posting) it is pretty unlikely that investigators will find locks or hindrances that may only be passed after solving puzzles. But this is something I will regard more thoughts with during play-testing.

4. Conclusion - summing up this thread!

Now, after those many words so far about AI-ing the investigators I think I am at a good point to test what I have come up so far. I now know the optimal "routes" through the Mansions of Madness; I also have an idea about how to make the investigators act during their turns; and besides this I have an idea about how to use skillpoints.

What I have not considered yet is - for example - in what order several investigators move during a single game-round. But here, I think, it makes total sense to keep the group close together; therefore I will focus on moving those investigators first that fell back the most.

Also I did not care about exploring rooms that are not "en route". Still they are "around" the major routes; so as long as I have more then one investigator in a game, I actually will modify and test the move-action slightly in the way that investigators may also move into rooms nearby (one wall away). This keeps the group close together and maybe earns the group important items.

5. For now ...!

Please! Feel free to comment on everything I have written so far. I will sum-up the results and post it on the BGG as well. I expect the BGG community to fire (and forget?) their comments on what I try here.

From now on I will playtest the results and - not unlikely - modify the results. About these changes I like to keep you updated. As soon as I have a "enough" solo-sessions tested, I will conclude the results in a final version.

All the best!

Mad

How is this solo project going, Mad?