For those of you who play X-Wing on Vassal, here are some important changes which are coming in the next 2 days. I explain the need for those changes. The most important one concerns the firing arcs, which were 90 degrees globally and will now be changed to 80 degrees, and just a tad too long. This forced turret weapons to be less good relatively compared to main weapons, even if gameplay was still balanced (but skewed) between the two players. If you think a -10 degree change isn't important, then talk to the Team Covenant community which, once it was alerted of the problem, would find instances of shots which shouldn't have happened in pretty much every game played on Vassal.

http://teamcovenant.com/mu0n/2013/11/28/the-road-to-4-6-0-and-better-firing-arcs/

edit - File download: (4.6.0 now available)

http://www.vassalengine.org/wiki/Module:Star_Wars:_X-Wing_Miniatures_Game

Not surprising as the chits were used as bases in vassal. I thought most people were aware of the discrepancy since the start? Personally I dont think its a big deal as precision on vassal is questionable whenever you are moving around pieces anyway.

Square shapes were used in vassal, so the intention was to create 1 graphic for both the plastic base and the chit combined. The mistake was to use the firing cones with lines that went to the corner of that global shape, creating a 90 degree angle.

I understand your disbelief that it doesn't matter much (it's unfortunately shared by a few), but through extended vassal play (which I don't know if you have or not), you can't help but notice "illegal", game-changing shots happen so often. It doesn't matter much in terms of game balance since both players have access to this, obviously, but it does matter for many people, myself included, who want an accurate representation of the game so that we don't acquire bad habits online and profit from situations we suddenly can't have in live games. I care about playing the same game in both versions.

Square shapes were used in vassal, so the intention was to create 1 graphic for both the plastic base and the chit combined. The mistake was to use the firing cones with lines that went to the corner of that global shape, creating a 90 degree angle.

I understand your disbelief that it doesn't matter much (it's unfortunately shared by a few), but through extended vassal play (which I don't know if you have or not), you can't help but notice "illegal", game-changing shots happen so often. It doesn't matter much in terms of game balance since both players have access to this, obviously, but it does matter for many people, myself included, who want an accurate representation of the game so that we don't acquire bad habits online and profit from situations we suddenly can't have in live games. I care about playing the same game in both versions.

Oh im sure it matters and that it allows illegal things to happen ... and its probably for the better that the changes go in to place. Im just saying that with the amount of printing variation and people fudging measurements in the actual board game, I do not think the average rate of erroneous (where by erroneous we mean things that were intended to be illegal in the game) was probably on par between the actual and the virtual games. So to play devils advocate, by making a hyper accurate version that does not take into account variability of manufacture, you are in fact creating something that is not representative of the table top and in turn you can get into situations on the tabletop that you can now no longer get in vassal.

It cuts both ways actually

In the creation of on graphic is where the error lies as the firing arch is not 90 degrees in the regular game. It is in fact unclear what the intended firing arch should be. My honest guess is that it *should* be 90, but because the plastic base *needs* to be larger than the chit to avoid spinning the chit (ie lip needs to be present), that the manufacturing realities affected the actually occurrence of the game.

Let's agree to disagree then:

-I didn't base my distance decisions on a single example of a ship chit, or a single range ruler. I actually averaged 9 small ships, 4 big ones using 1200 dpi scans, and painstakingly measured the distance with range tools in gimp, calculating the whole thing in excel. My process can be found here: http://teamcovenant.com/mu0n/2013/11/11/definitive-correction-to-the-vassal-firing-arcs/

Your argument that I might have botched all this new work due to printing error is invalid. I measured an average and decided upon sound rounded intended values for the simple lengths (ship sides) and an accurate as possible value for the most important new corrected length, the cone length on the side of the chit. I *EVEN* provide a stats based uncertainty for my value. If you can provide a better way to do all this AND avoid manufacturing variance like I did, I'm all ears.

-I think the intention was an angle clearly below 90 degrees, as per the rules which suggest we use the range ruler side and line it up to the printed firing cone lines

-Errors in ship manipulation in live games go both ways (undershoot and overshoot movement), so the long term average of all of these approaches close to the intention. Meanwhile, in vassal, it was 100% of the time overshooting movement. That's pretty bad in my opnion.

edit - Vassal needs integer values for movement, despite the distance ratios inevitably leading to decimal point values (banking, turning, arcs, etc). Vassal isn't hyper-accurate. Ship collisions still demand manual placement + visual judgment and will be fudged too. Your other argument isn't accurate as well.

Before you start saying that these vassal errors make all of my tedious work completely useless, let me pre-counter that by saying the biggest arc and movement mistakes were a good 4-7 pixels off. The edge of the firing arc that was too large can fit close to 3/4ths of a small ship base. Those errors are huge and not within an acceptable +/- 1 pixel error. They did matter and did lead, many times, to situations which shouldn't have occured. I'm not a solo lunatic getting worked up on nothing, I have many fellow heavy vassal players which will be happy to confirm my sayings.

whats your sample size? 95% confidence interval is fine in general, but rather meaningless if your sample size is low and not representative of the manufacturing process. Also, using an average is NOT the way to go.

You have not provided enough evidence to convince me that your approach is in fact valid. Show me a CDF of a reasonable sample size and pick your degree for the firing arcs from that, don't use an average. Think of writing a technical paper, this reviewer does not buy it.\

Basically, I do not find your process correct and disagree with how you went about it. However, in general it's probably more correct than what was there before (though we do not have convincing statistics to make that claim either).

Zreef chill out and thank this man for his time and effort put into building up the community. If you go over to the Team Covenant website you will see many discussions on these issues and how the community as a whole felt they should be handled. Mu0n is simply the one pushing the hardest and putting the most into the game.

I was told my argument was "invalid", im simply showing that it is in fact very valid.

If you are obsessed with precision and representing things as they are on the tabletop, as was claimed -- there is a correct and an incorrect way of doing it.

Now im perfectly willing to accept that this is a "good enough" solution, but claiming it more accurately represents the table top is not correct. I object to the claims being made.

The process is fine and dandy, but does not support the claims. Is the solution "good enough", I am sure it is. The bigger question is was it necessary? That, we do not have quantitative support for, we have, for better or worse, opinions.

here is what you can do:

1) gather a sample size that is large enough, say 1,000 measurements per unique chit (probably not statistically significant of an amount, but lets keep this do able). To do this you will probably need to crowdsource the measurements (so we will have to just assume our measurements are reasonable)

2) create a CDF of these measurements

3) modify the overlay in vassal to take as an input the CDF and generate the overlay uniquely per ship per session. The vassal engine would support this, but would take some book keeping.

Now you get an much more precise representation of the table top, that takes into account the variations in the production process of the game. Picking a single angle, regardless of how the process was undertaken to get it, is not going to give you a more "accurate game" if you measurement of accuracy is the table top.

Check section VI of this document from a reputable (though you might disagree?) academic institution, Berkley. This is what I used. A confidence uncertainty rooted with t of Student distribution. http://www.cs.berkeley.edu/~brewer/cs262b/Lec1-stats.pdf

If you bothered to check my links above, you'd see I have a 8 small ship sized sample, so let's drop that to 7 degrees of liberty. I have a custom program which calculates the uncertainty of the average value. You multiply a coefficient (is an atrocious process to generate them for custom confidence level and custom sample size, but I programmed all of that back in 2011) with the standard deviation of the mean. My generated number (7 degrees of liberty) is 2.36526023122995 which matches what can be found in the pdf above.

In my opinion, all of this process is COMPLETELY OVERBOARD for the needs of fixing some firing arc in some game, but I happen to be able to do it. I happen to enjoy doing it, despite being upwards of 40-50 hours of unpaid work, purely out of my love for this game and the need to build a strong online community of vassal players. It practiced what I learned during my physics B.Sc. and is part of some of the things I teach to 17-18 year olds. I haven't been (surprisingly) been exposed to CDF distribution function for this kind of purpose. If you're willing to show me how it's used and how much better it is to get data from it, be my guest, I'll be interested. My simplified rule of thumb for stats is to use a normal (gauss) distribution when you have 30+ samples and to use t of Student when you're under it. Anything more complex that that goes way beyond the level I teach at normally. It goes even more beyond what's needed here.

You came in pretty hostile in the thread completely negating the need for precision and dumping on my work, then require better university level statistics. I'll leave that behavior up to public judgment.

The difference is whether or not you want a single measurement for the arc or not.

I did not dump on your work, I questioned whether or not it was necessary. Then you stated it was more precise, then I questioned your process and your definition of precise. I will shoot you an PM and explain my thoughts on the matter, one educator to another. As this is quickly going down the rat hole and I don't want to play the degree game.

So I will apologize for my hostility, it was not intended to be hostile.

If it helps mu0n, a simple Thx from me for your effort.

Never noticed that but that makes a lot of sense as the Chits aren't square. Thanks for the update. And Mu0n we will talk after the holidays, I'm still willing to help out.

From someone that plays on Vassal I appreciate all you do for the community Mu0n. I do not get to play very often on Vassal any longer because I have been busy with stuff in life right now.

There always has to be someone that wants to critize instead of thank you, or help you for all the hours you spend doing something for FREE for the whole community. I see that he has not been keeping up with all the work you have been doing, and just likes to discredit where credit is due.

Lets see if he would be willing to take his time for FREE and teach us all CDF distribution. To do that HE will need to take the 1k measurements and croud source HIS research to prove to the rest of us HIS work.

You have showed us what you have done allready Mu0n.

@Grumium, you are correct. I do not have the time to take up the process I outlined. If you are interested in why I objected to the claims, shoot me a PM.

If you are interested in CDFs, wikipedia has a reasonable overview: http://en.wikipedia.org/wiki/Cumulative_distribution_function

I have sent MuOn a PM and apologized to him directly, as the discussion was heading in a direction that was not constructive.

I accept your apology, zreef. The good that came out of it (for me) is that I become aware of CDF methods in a cumulative measuring process, which is cool it exists. It might help me in the future (though not for x-wing). It really comes down to a global philosophical approach of wanting to play a simulated game based on unchanging ideal measures, or a real-life variance simulation on the computer. I have never heard this being done for any tabletop game ported to a computer game before. In my mind, they use static values for lengths every time.

Since there's a new online tournament starting next week over at Team Covenant, it's auto-out of the question to go there and do this approach. I don't think that even with the time, I'd go with this unprecedented approach that adds a bit of randomness to every game, I guess? I mean, we already have dice providing ample random factor. If you were able to prove that my particular data sample made me undershoot or overshoot by something like 3 pixels every time, then I'd take pause and humbly accept any correction. My gut feeling is that this proof will never be made anyway (I hope no one cares enough to do it, it's just a game).

"It really comes down to a global philosophical approach of wanting to play a simulated game based on unchanging ideal measures, or a real-life variance simulation on the computer. I have never heard this being done for any tabletop game ported to a computer game before. In my mind, they use static values for lengths every time."

This is exactly what I was banging around in my head when I read your post and the data you presented. In retrospect it was poor taste on my part to argue this in the thread where you announced your work. So once again apologies to you and all.

"The good that came out of it (for me) is that I become aware of CDF methods in a cumulative measuring process, which is cool it exists. It might help me in the future (though not for x-wing)."

Its a very useful method and I am glad that something good came of it.

"I don't think that even with the time, I'd go with this unprecedented approach that adds a bit of randomness to every game, I guess?"

I do not think I would go with it either, too much hassle for an unknown benefit. I could see it being useful for people who want to play games professionally, for everyone else it might be overkill.

"If you were able to prove that my particular data sample made me undershoot or overshoot by something like 3 pixels either way, then I'd take pause and humbly accept any correction."

Its unclear either way, we would need a lot of measurements to know for sure.

Thanks! Great update.

I believe I know what you mean zreef. If the paper squares are not all 100 percent accurate then the end result will be different. The tip to tip at the cone section will be off.

I am not interested in taking 1k measurements. The CDF is a little more advanced then what I am into right now, but thanks for posting the link.

Thanks for apologizing to Mu0n. He has done alot of work for the Vassal community and deserves a thanks for all his work.

Thanks for all your hard work MuOn!

See-through functionality for ships through a toggle shortcut: CTRL-G (G for Ghost), has been found (thank you Theorist!). This is now necessary since the ship bases are no longer transparent. You need to see under the ship while moving it around on a template, when ship to ship collisions happen.

This is great work, Mu0n. I can't wait to play with the new models in Theorist's VASSAL tournament.

Paul

Cool - thanks!

I need to check on a tourney as well...

Some people have trouble installing the new version. Here is a simple fuss free guide:

http://teamcovenant.com/mu0n/2013/12/02/need-help-installing-v4-6-0-without-fuss-even-on-top-of-an-older-version/