11 minutes ago, jimmius said:

**** them then. they can pump out these hideous new ships all they like, but theres no excuse for the travesty that is the X-Wing in it's current state

You need to chill out. X-Wing is not a travesty...its a toy spaceship game.

Complaints and discussion about issues with the game are normal and welcome. Just discuss those issues like a rational person.

This community is better than crap posts like you just made.

Criticism comes with the territory but attacks like the first sentence are not cool man.

A neural net for comparing lists would be a cool thing I could get behind, especially if it could receive the data from tournament results as well as crowdsourced opinions.

I still have my doubts about the ease of creating a functional x-wing AI though.

Making one that can pick reasonable moves and modify dice efficiently is relatively easy (I mean, Benchmark has a generally OK one for that, but it has no skill and occasionally commits suicide), but making one which can do the kind of heuristic assessment that the mk 1 brain can is really not trivial.

5 hours ago, Azrapse said:

@MajorJuggler I bet you know of the X-wing Squadron Benchmark and Fly Casual projects. These project let you play against an AI that is just basic at this moment.
I know AI is not your area of expertise but, do you think developing a competent AI that can play a whole match against itself in matter of a second or two (simulating human times for movements, dial planning, action taking, bump resolving, in order to limit virtual playtime to 75 minutes) would help analyzing how efficient ships are?
I mean, if we could define every ship and upgrade card they way Benchmark or Fly Casual do, but so that the computer can run a game without visual interface, all silent number crunching, and play the entire match in seconds, then retry many times gathering some statistics of how every ship perform, over and over again for many thousands of matches, would it be any useful even if we don't get an AI that is as brilliant as a human? Would that help figuring out what ships/cards are just better regardless of player skill?
Or would that be greatly undermined by the limitations of the AI skill level?

Writing an AI that can play the game intelligently would be the cornerstone of the holy grail of automation. Everything starts there. Once you have a machine that can play the game against itself, and play well, then you can leverage that for a lot of different things:

• Seeing how effective certain lists are at other lists
• Backing into those results to get an indirect measurement of how good certain ships are

There's a few ways to go about this. Conventional wisdom is to write the AI code yourself and hope it's good enough. But a better way is to just feed the rules to a Neural Network, give it 0 strategy, and let it play against itself 100,000 times, to let it self-train the optimal strategies. This is the approach that the current Go Neural Network used, and it managed to beat the previous bot (that had dethroned the humans) quite handily.

Once you have a bot that can play the game, then the real fun begins. Then you can create a AI / Neural Network layer at the list building stage, which can try building new lists and find new combinations. You would use the first bot to test the lists and see how feasible they are.

1 minute ago, Boom Owl said:

You need to chill out. X-Wing is not a travesty...its a toy spaceship game.

Complaints and discussion about issues with the game are normal and welcome. Just discuss those issues like a rational person.

This community is better than crap posts like you just made.

Criticism comes with the territory but attacks like the first sentence are not cool man.

they've had literally years of data proving the X-Wing is a terrible ship, and they've decided to give up with a pithy "we're looking to the future not the past". If anything I'm being polite

2 minutes ago, jimmius said:

they've had literally years of data proving the X-Wing is a terrible ship, and they've decided to give up with a pithy "we're looking to the future not the past". If anything I'm being polite

Thats fine. Talk about that then we all do. But don't say **** em. This community should be better than that crap.

49 minutes ago, jimmius said:

so I don't have time to listen to this, not really been into X-Wing for years (on strike until they fix the T-65)

any word on that front? Last news we had was from 2015 when Alex said "IA is not the last we'll see for the Old X-Wing".

Or are we stuck with revolting turdlike spaceships forever until the game dies?

You seem to miss the idea that other people like what you describe as "turd like spaceships". There are plenty of EU ships I really want and would probably still fly more than a fixed X-Wing.

You don't have to like the other ships, but at least have the courtesy to acknowledge that others don't hold that same dislike and that's fine.

Just call the game " Spaceships" .

11 minutes ago, Velvetelvis said:

Just call the game " Spaceships" .

OR!

Star Wars: A space based Miniatures Game!

Star Wars Legion: A ground based Miniatures Game!

Star Wars Destiny: A dice card based game!

Star Wars TOT: A game only featuring seven ships!

Gunboat: A Star Wing style XG1 Gunboat game of gunboats set in the Star Wars Universe where Gunboats once roamed. Art by Bob Ross, narrative by SIRI. Manual written by Clippy of Microsoft.

44 minutes ago, Makaze said:

I completely disagree. While the number of possible board states is quite large the number of available moves on any given turn is relatively small, especially with some lists. Making something that plays fairly well while looking forward 2-3 turns would not be overly difficult.

There's also the question of who it's supposed to beat. I once, in 48 hours, wrote a chess program that can reliably beat me (1 minute turn timer on a single core 212Mhz Pentium 1). It basically just did its best not to make bad moves and apparently, as a chess player, I personally don't meet that standard. Taking a similar approach with X-wing I have little doubt you could write something that plays around as competently as the average FLGS player just by maximizing its shots and minimizing the opponents shots. Even if it would likely get spanked by higher tier tournament players.

Chess is super-easy to program for compared to something like X-Wing. You've got a set beginning game state and a limited number possible moves per piece, plus one move per player per turn.

Go is orders of magnitude more difficult than Chess but it still has a fixed start point and relatively simple rules, while a customisable game is orders of magnitude more difficult than Go.

6 minutes ago, Stay On The Leader said:

Chess is super-easy to program for compared to something like X-Wing. You've got a set beginning game state and a limited number possible moves per piece, plus one move per player per turn.

Go is orders of magnitude more difficult than Chess but it still has a fixed start point and relatively simple rules, while a customisable game is orders of magnitude more difficult than Go.

Succinctly and well put.

My main association with the difficulties of AI development was in an open source turn based strategy game, and this was basically the logic used there. The FAQ was ;surely it shoudl be easy to develop an AI for the game' and the answer was 'lol no'.

It would be relatively easy to develop an AI for one specific game of x-wing - i.e. two fixed lists with a fixed opening board state and fixed starting positions - by using fairly standard neural net programming. But developing an AI that could reliably play ANY list effectively, right from setup, AGAINST any list... that would be the trick.

1 minute ago, Stay On The Leader said:

Chess is super-easy to program for compared to something like X-Wing. You've got a set beginning game state and a limited number possible moves per piece, plus one move per player per turn.

Go is orders of magnitude more difficult than Chess but it still has a fixed start point and relatively simple rules, while a customisable game is orders of magnitude more difficult than Go.

But far fewer pieces and in many situations the number of reasonable moves on a dial (ie. don't fly off the board, don't land on a rock) are no more than the number of squares non-pawn chess pieces have access to once the board has opened up. The number of static evaluations of branching board states is not appreciably higher than a mid-game chess board. Openings, just like chess are easily prescripted. I've no doubt that a single turn of X-wing has more decisions but there are fewer turns. So even if it is more complex overall we're still in the same ballpark.

Go is, to be honest, also not an overly hard game to write an AI for, at least if your goal is only to beat an average player. I'm not talking about building the AlphaGo or AlphaGo Zero of X-wing, that would be an incredible challenge. Though I'd also argue an unnecessary one as I don't think the best human X-wing player is anywhere close to as good as the best human Go player in terms of relative study, depth of understanding, and knowledge of the game. I'm talking the equivalent skill level of a free chess/go program you to play on the bus on your phone.

Biggest challenge I see would be how to deal with the random elements. If you shoot a ship there's a chance it goes away and you can disregard it in future trees, but there's also a chance it doesn't. Do you run out a weighted tree for every possible damage distribution or do you give things psuedo hp ranges that represent a couple standard deviations of what hps they "should" have in that state? That's actually an interesting question, one blows up your state count immensely while the other complicates your evaluation...

49 minutes ago, Viktus106 said:

Gunboat: A Star Wing style XG1 Gunboat game of gunboats set in the Star Wars Universe where Gunboats once roamed. Art by Bob Ross, narrative by SIRI. Manual written by Clippy of Microsoft.

This would obviously be the best x-wing spinoff...

There has been a LOT of talk about game balance, and competitive meta, and how the designers aren't doing that great choices. I just want to add that I think for the majority of pilots and upgrade cards that are designed, the number one question the developers ask themselves is, "Is this fun to play?" You really get a sense for it towards the end of the interview (page 21 on the transcript) when they start talking about what cards they think are underrepresented. You can feel their excitement start to come to the surface.

I think - and this is coming from somebody that doesn't take tournaments all that seriously (I fly whatever I want!) - that their biggest mistake was trying to design cards to counter the meta. The "meta game" they need to look at is all the cards that they've produced. Like others have said, creating tech against a single problem card/combo can be problematic if the evolving game ends up deemphasizing its strong points, and it fades into the background. I think they should just keep trying to come out with fun cards, and some will be a hit with the competitive crowd, while others will not. I think expecting this game to be completely balanced at a competitive level is asking for a bit much.

Now, the Community Mod that will be coming out will be an excellent way to play with access to all the cards, and may be a better competitive game. I'm looking forward to it. I just don't see how FFG can course correct this game at this stage in its life, and I think it's unfair for us to expect that from them.

P.S. If they didn't take the meta into consideration when designing new cards, I don't know if Autothrusters would have been a thing. I wonder what a "no Autothrusters" game would look like. My TIE interceptors would be sad, but I'm also perplexed that these ships' weakness has since become their strength.

Re: simulator, there have been a couple of attempts at this:

• the more recent Fly Casual sim (github: https://github.com/Sandrem/FlyCasual )
• the older Squadron Benchmark (https://xws-bench.github.io/bench/)

In terms of the technical challenge of implementing an actual player:

• As with Alpha Go and Alpha Go Zero, a reinforced neural network is the place to start . Basically the software is its own teacher, playing multiple games against itself to incrementally improve. (Note in the first version of the program, it actually first trained using master games and human's showing it some of the basic techniques in Go -- the newer one started from scratch.)
• As mentioned upstream, X-Wing has more hair in terms of rules and game phases, but is tractable. The X-Wing Data repo would be enriched to include json that declaratively injects the rules system into the simulator. X-Wing has way more corner cases that a game like Go or Chess, and it is much more flexible to solve those cases in an expert-managed json rules repository than in the actual source code :-)
• Simulation then begins in a "headless" fashion -- all of the simulation runs happen out in a big cloud compute cluster, without a human being or monitor connected to it. You could put something like FlyCasual on top of it to give us dumb humans a chance to observe and enjoy, but that isn't really required.
• Once the RNN has trained itself sufficiently, you must then deploy that model to go beat the crap of actual human beings. For example, you could inject the model into Fly Casual and then play against it!

All pretty do-able stuff, but that's a real software project, where you take two senior programmers and one data scientist and six months and expect them to come back and tell you that it'll be another six months. Kickstarter, anyone?

As an aside, for many years the AI crowd was hotly anticipating the moment when AI would "solve" Go. It was a really big deal, phase one of a larger end the world operation (phase two: AI starts stealing underwear from your wife's panties drawer, phase three: global nuclear destruction). Now its happened, and Google actually open sourced the software library they used to create it! (Tensorflow). Phase two isn't far now ;-)

And I would say it is unfair for them to continue to churn out substandard product (now with some willful ignorance) and expect the consumer to keep buying and increase the cost of doing so WHILE devaluing prior purchases.

It’s hard to convince others to play “alternate formats” when they just spent $$$ on the dominant format. Even harder when they are new and don’t have access to all the “fun” cards from prior waves because they specifically didn’t buy it because it’s obselete.

Also, FUN for me is a balanced game. Xwing is not there right now.

Quote

RW: So how did that ship come from the...how did you guys decide, or how did you get that in front of you to design it? Did you have to say “guys, there’s a 150 page request for this ship?” Does that come into play, or how does that work?

Frank: We didn’t actually have a lot of control over what the ship’s were even suggested to play with for a long time, because a lot of those decisions were being made above us. So obviously they wanted to incorporate a lot of ships that were in the movies first. So they were like, yeah we’re definitely going to do the interceptor, we’re going to do the bomber, etc. So it took us a pretty long time before we were actually able to make suggestions about what ships were being added to the game, so on some level it didn’t matter how much the forums were just shouting, the people making the decision were unaware entirely.

(snip)

Alex: It was our plan all along. So yeah, as I recall, I leaned over to Frank one day while I was one of us was putting together documents to send to somebody which was concepts for new ships, I think I said to him but maybe he said to me “It’s time”

Frank: It was you. It was definitely you.

Alex: And I seem to recall that you were like “not it’s not time yet” and I was like “No, it’s time.” And so we put it on the list not really thinking much about it, and then it came back and they were like let’s do this was. And we wanted to see what LFL has to say about it, because of course everything we do has to go through them and they often have good suggestions. Don’t do this ship, we’re about to put a book out that will invalidate everything about that ship.

Tyler: Strong.

Alex: In this case, they said no, go for it. Looks cool. They were actually pretty excited.

I want to go back to the (unrelated) point I made some pages back, about the dastardly House of the Mouse and its iron grip on all things X-Wing. Because now I can cite textual support

The Evil Mouse of Evilness wasn't driving the U-Wing, the Ghost, etc. An unnamed individual within FFG was (and, real talk, it makes sense. Of course you want to release a movie or TV tie-in as close to the point of maximum interest as possible. Plus, if you're an active fan of those properties (and they seem to be), you're more actively excited about ongoing and upcoming products than your memories of Star Wars Galaxies Of course you want the Interceptor before the Defender. Of course you want the Ghost and its crew before the gilamonster).

Like, there's room to think otherwise. The exchange isn't completely cut and dry. But I'm willing to say that the implication is pretty clear.

It's very interesting to note the significant changes made in the Destiny FAQ. Among meta shifting rule changes are also point changes for cards. We've seen it in IA but now that's two SW properties, so I'm wondering now who is ultimately responsible for keeping it out of Xwing (though maybe a new FAQ will change that).

Hey there folks, it's me again, just checking in again to let you know that I've tidied up and completed the list of links in the alternate formats thread. Further suggestions are always welcome as we strive for the distant goal of 300+ pages to get attention from the devs like they mentioned in the podcast. Unfortunately, the alternate formats thread isn't powered by a salt battery and so it's lagging behind this one a little. If you have any alternate format ideas, critiques, hopes or fears, please contribute to the alternate formats thread!

On 2017-10-16 at 4:49 PM, Kieransi said:

I get what you mean... it’s hard to root for the Empire. I just think everyone benefits from a fair and balanced game.

I think you get weird looks from people if you play Axis and Allies and root for Nazi Germany, but it’s no fun if the allies win on turn 1, right?

Besides, I don’t really want to get into this debate, but I think pirates are still on about equal moral footing with nazis (walking the plank never happened in real life - Hollywood invented it because they aren’t allowed to show what pirates really did to prisoners).

They did show it in Black Sails and it was ugly...

21 hours ago, AngryAlbatross said:

This would obviously be the best x-wing spinoff...

I heard it's getting its own spinoff film.

Gunboat: A Star Wars Story

It's a lot like Saving Private Ryan but Ryan is actually a prototype Gunboat (OS version) and the team are a combination TIE F/O and new TIE/SFO bombers tasked with retrieving the last Gunboat because losing it would be devastating to Imperial morale.

Instead of that famous scene of Tom Hanks valiantly destroying a tank with a pistol, it's actually our protagonist, in the Gunboat suffering the effects of a overused SLAM drive, forced to fire it's primary weapons at a squadron of Jumpmaster 5k bearing down on it.

Just when all hope is lost, the target lock warning light frantically beeping, shields depleted and communicated lost. . the JM5k explode in a spectacular display of Imperial retribution as THE NEW Star Wing 2.0 F/O (which looks exactly the same, but with a blood stripe now) drops out of SLAM and blackens the sky with precision ordinance. The resulting explosions last longer than the head burst scene from Kingsman and it's a true testament to fandom.

First off, kudos to Team Covenant and to the devs for the interview, and to Ablazoned for the transcript. This is a great tradition, and please keep it coming!

My 2c:

Current state and the future of the game:
I vote "no confidence" based on what I've heard on the interview.

Based on how things are setup at the moment (overall funding level for XWM, size of the design/test team and their focus/priorities, their skill set and tools/processes they are using, their design goals: optimize for "fun" and "making new things good so people buy them", etc.), I think its is very unlikely that this game will move to a "more balanced" state - quite the opposite. FFG will very likely stay on the "current course" as long as the game is profitable for them, which is definitely the case today (and will likely continue to be the case for a while going forward). If/when the game stops being (very) profitable, they may change things more drastically (re-balance the rules and/or create a 2nd ed.). But that may not happen soon (if ever).

Based on this thread (and other threads on this forum), it seems that there is a large number of players out there that are unhappy with this (myself included) and are no longer enjoying the game as much (or at all) due to the poor overall balance. However, our feedback is either not coming through, or it's not deemed important enough to outweigh the (likely increasing) revenue coming from the other players. So FFG may choose not to react until it's too late.

What we can do about it:
It seems that our best shot in getting a more balanced game (where more ships and more playstyles are viable and balanced, incl. the older releases) is to either:

1) Help the designers / FFG in some way (with new/better community-driven tools, better testing support/process, ...)

2) Create a community version of the game (a la "The 9th Age" for Warhammer Fantasy) with more balanced ruleset

3) Vote with our wallets and look for greener pastures (and apply "financial pressure" to FFG to change their policy)

What do you guys think?

P.S. Do we get a bonus if we get this thread to 300 pages?

40 minutes ago, vladamex said:

2) Create a community version of the game (a la "The 9th Age" for Warhammer Fantasy) with more balanced ruleset

Working on it! X-wing Community Mod is currently in closed alpha. Although, I am getting a very limted amount of data, and I really need my first "wave" of ships playtested, so I might open it up soon.

43 minutes ago, vladamex said:

3) Vote with our wallets and look for greener pastures (and apply "financial pressure" to FFG to change their policy)

This is my current plan. I will probably buy some Gunboats (for nostalgia) and the new ghost shuttle (IF flight astromech is pretty good). Otherwise I don't see anything that interests me in the Kimogila, Silencer or Resistance Bomber.

1 hour ago, Viktus106 said:

I heard it's getting its own spinoff film.

Gunboat: A Star Wars Story

It's a lot like Saving Private Ryan but Ryan is actually a prototype Gunboat (OS version) and the team are a combination TIE F/O and new TIE/SFO bombers tasked with retrieving the last Gunboat because losing it would be devastating to Imperial morale.

Instead of that famous scene of Tom Hanks valiantly destroying a tank with a pistol, it's actually our protagonist, in the Gunboat suffering the effects of a overused SLAM drive, forced to fire it's primary weapons at a squadron of Jumpmaster 5k bearing down on it.

Just when all hope is lost, the target lock warning light frantically beeping, shields depleted and communicated lost. . the JM5k explode in a spectacular display of Imperial retribution as THE NEW Star Wing 2.0 F/O (which looks exactly the same, but with a blood stripe now) drops out of SLAM and blackens the sky with precision ordinance. The resulting explosions last longer than the head burst scene from Kingsman and it's a true testament to fandom.

OH COME ON DUDE

at least put a spoiler alert in there