[runabout]

Still thinking "out loud"

I saw this post on the forums about solo ST: viewtopic.php?p=165398#p165398

It looks like that solo mode takes a lot of time to set up, but it made me think about what could be possible.

At some point Decipher added "scenarios" to Star Wars CCG. You had to reach a certain number of points by playing certain cards and performing certain actions. These were not difficult to adapt to play solo, and it was easy to create new scenarios. There are some examples here (for the normal, 2-player game): http://theforce.net/ccg/sw/scenarios/

Would that be a possibility in Star Trek CCG? The dilemmas would still be random, but if you decide at the beginning which dilemmas or which type of dilemmas you have to face it might not matter as much.

[frakkingoff]

But .. a few "weeks" of solid coding? Really? I don't know anything about coding but that seems like something that would take at least a 6 months no? To get the AI to not be completely stupid would be tough, no?

Well, by a few weeks of coding, I mean working on the algorithm full-time (rather than as a hobby project). And I suppose I meant for someone experienced in such AI algorithms - I might be able to get a rudimentary version online in a few weeks, but I've been working in AI simulation algorithms recently. A non-AI programmer might take another month or two to delve in first.

The beauty of adversarial networks is that, once they're running, they teach themselves. You give them the framework, but then you stand back and let them learn on their own. So a few weeks to setup the initial code, and then let it run in the background (rather than old-school AI, where you have to program all the decision-making and rules that the AI follows - adversarial networks can learn what the "right" decision is by itself). Come back a few weeks after that and hopefully you've got something.

Also it helps that the card database from this site is probably easily ingested (so you don't have to hard-code the text of all the cards yourself - though the dilemmas will probably have to be "translated", which might take another week or two).

At some point Decipher added "scenarios" to Star Wars CCG. You had to reach a certain number of points by playing certain cards and performing certain actions. These were not difficult to adapt to play solo, and it was easy to create new scenarios. There are some examples here (for the normal, 2-player game): http://theforce.net/ccg/sw/scenarios/

Hey, that looks cool, I never saw that before.

Would that be a possibility in Star Trek CCG? The dilemmas would still be random, but if you decide at the beginning which dilemmas or which type of dilemmas you have to face it might not matter as much.

Well, assuming you solve the dilemma problem, then there's no reason you can't play solo by trying to rack 100 points (or whatever total) in as few turns as possible. The quicker, the better - you can challenge yourself to make better decks (or riskier gameplay decisions) to beat your "high score".

[runabout]

Hey, that looks cool, I never saw that before.

Yeah, it was a lot of fun. Sadly, I don't have those cards anymore

[Se7enofMine]

But .. a few "weeks" of solid coding? Really? I don't know anything about coding but that seems like something that would take at least a 6 months no? To get the AI to not be completely stupid would be tough, no?

Well, by a few weeks of coding, I mean working on the algorithm full-time (rather than as a hobby project). And I suppose I meant for someone experienced in such AI algorithms - I might be able to get a rudimentary version online in a few weeks, but I've been working in AI simulation algorithms recently. A non-AI programmer might take another month or two to delve in first.

The beauty of adversarial networks is that, once they're running, they teach themselves. You give them the framework, but then you stand back and let them learn on their own. So a few weeks to setup the initial code, and then let it run in the background (rather than old-school AI, where you have to program all the decision-making and rules that the AI follows - adversarial networks can learn what the "right" decision is by itself). Come back a few weeks after that and hopefully you've got something.

Also it helps that the card database from this site is probably easily ingested (so you don't have to hard-code the text of all the cards yourself - though the dilemmas will probably have to be "translated", which might take another week or two).

At some point Decipher added "scenarios" to Star Wars CCG. You had to reach a certain number of points by playing certain cards and performing certain actions. These were not difficult to adapt to play solo, and it was easy to create new scenarios. There are some examples here (for the normal, 2-player game): http://theforce.net/ccg/sw/scenarios/

Hey, that looks cool, I never saw that before.

Would that be a possibility in Star Trek CCG? The dilemmas would still be random, but if you decide at the beginning which dilemmas or which type of dilemmas you have to face it might not matter as much.

Well, assuming you solve the dilemma problem, then there's no reason you can't play solo by trying to rack 100 points (or whatever total) in as few turns as possible. The quicker, the better - you can challenge yourself to make better decks (or riskier gameplay decisions) to beat your "high score".

Yeah I guess the whole concept of algorithms just seems so .. big .. to me. I can't warp my head around it so in my mind, it seems like it would take forever to do.

I nominate you to get this rolling. Muahaha.

Do I have a second vote? Anyone? Anyone? BUELLER? Heh

[CaffeineAddict]

Ive dabbled with playing solo variants, and I find the key thing is that you need to be able to lose. If the game doesnt have a way to beat you, then the game is just "how long till i win".
Ive tried things like limiting the number of turns, time limits, and having the number of points needed to win increase each turn - so the longer it takes, the harder it is to win, and when the needed points hits a certain number, you lose.

A variant ive tried a few times is starting out with a ship and senior staff, and supplementing from your draw deck throughout the game. Then I increase the number of dilemmas per mission to 1 per 5 points its worth, sometimes with an additional couple as well. Add in the increasing point requirements, and it can be quite challenging.

[stcharlesmatt]

Did anyone get anywhere with this idea? I think this would be the best shot at solo rules! If not, does anyone have a list of the top combos?

had a thought... it would be cool if the community created a sort of "combo database" say of 1000 combos they have used or seen used in their lifetimes. Then we have a QR code that randomly "dials up" a combo out of the 1000 similar to that doorway Quantum Incursions. That way you can simulate actually facing a real person whose dilemmas you don't know. "Combo 154: Lack of prep + the higher the fewer + cytherians" or some such

Drawing random dilemmas really doesn't help you much. it is real life combos you want to face.

That would be a lot of work! It would only work on such a scale, because if you did it with fewer cards this means that, after a few games, once you've seen the first dilemma you know exactly what the others will be and you can prepare for them appropriately. That will also take away the surprise that comes in a normal game.

[Tim]

Some thoughts:

In the olden days, combos wheren't 6 different combos but more of an over-all denial strategy. Since OTF 2 rule that's been less so. Dilemma combos would be separate? Or would you want to emulate a real player, with 6 combos always appearing together?

While abstractly dilemma combos are an interesting research paper idea on machine learning, in all practicality just a database of combos would probably work better.

Would combos be indexed to specific missions, or does it make sense to give combos broader rules? Or maybe allow both?

[Professor Scott]

When I play solo, its generally playtesting my decks against each other. There is no getting around knowing what combos I am playing. This is actually somewhat similar to knowing the Meta for the area I am playing in.

Here is a suggestion to help:

After placing missions, go ahead and seed dilemmas as you normally would under opponents missions.

Then stop. Go do something else for an hour.

Then come back and seed under your own missions and proceed as normal. This gives you a chance to forget exactly where the dilemmas went so now you have a more realistic uncertainty.

For the record, I love the idea of a Quantum Incursionsesque dilemma combo seed card.

[Tim]

Thinking about a data structure for dilemma combos, and I'm back at this question:

How important is it that a set of 6 combos appear together? Or good combo sets still overall denial strategies?

[pfti]

Thinking about a data structure for dilemma combos, and I'm back at this question:

How important is it that a set of 6 combos appear together? Or good combo sets still overall denial strategies?

I play combos that are adjustable based on the spaceline. So it is built usually around 2 card combos and flexible dilemmas to go with them

[dragoncymru]

I've been playing around with a solo version that takes the game back to the roots and STNG!

Designed as a system to make a game emulate a season of TNG rather than a deck-building or competitive experience, it is working out quite well.

And you get to have Away Teams of Riker, Data, Worf and Geordi, rather than 3rd Ensign from a show you don't remember!

I'm not sure whether people are interested in seeing the ruleset or even the card database I've created for STTNG Seasons 1-2 (and part of 3) but if people want to see it, and tell me the best place to publish, I'm very happy to share and see what people think?

let me know?

[Professor Scott]

I've been playing around with a solo version that takes the game back to the roots and STNG!

Designed as a system to make a game emulate a season of TNG rather than a deck-building or competitive experience, it is working out quite well.

And you get to have Away Teams of Riker, Data, Worf and Geordi, rather than 3rd Ensign from a show you don't remember!

I'm not sure whether people are interested in seeing the ruleset or even the card database I've created for STTNG Seasons 1-2 (and part of 3) but if people want to see it, and tell me the best place to publish, I'm very happy to share and see what people think?

let me know?

This is very intriguing to me as I seldom get to play with others and mostly just playtest against myself.

[dragoncymru]

Nice to hear, as I said, if you show/tell me where to post the rules and card-database I'll let you see them.

Still a WIP mind, and slight polishes being done all the time as I playtest, but it's good fun nevertheless and emulates a TV episode well.

Here's a Mission report, for example, of a turn in a game last yesterday:

Picard, Guinan, Data and Worf beamed down to investigate an anachronism detected on a planet. The team discovered a Clone Machine before Data was captured by Pakleds, Picard was killed by an unknown virus and the Away Team was stopped.

Fortunately, the team was joined by Donald Varley who arrived in the USS Yamato and Kate Pulaski who beamed down to the planet. The combined cunning of Guinan, Worf, Varley and Pulaski was enough to rescue Data from the Pakleds before they encountered alien child Sarjenka, who they (naturally) saved and were stopped again.

The final part of the Mission saw Pulaski agonise over using the illegal Cloning Machine (and lose 5 points) before resurrecting Picard.

Now if that doesn't sound like a TV episode....

[Boffo97]

We don't play 1E because it makes stories that sound like episodes... we play 1E because it makes stories that sound like fan fiction written under the influence of illegal medication.

[dragoncymru]

Each to his own!