4606

I was looking forward to write about command and control, delegating stuff to the AI etc. rather than the manual, but as I translated a good chunk of the original WITE manual into Spanish as part of a community effort, I can give you some informed commentary on that.

4607

Would love to hear!

4608

The problem here is that documentation is substituting for clear game design in favor of a perceived deep simulation. And given the amount of fudge WitE had to do to make the Soviet comeback competitive I have doubts about the simulation quality of such systems anyway. That perceived deep simulation is in many cases nothing else but smoke and mirrors that produces outcomes that could be had with a simpler approach. Window dressing that adds immersion to the game but with no real mechanical benefit (and I do like WitE). Fidelity as texture and not substance.

I vastly prefer clear game design to obfuscating detail supported by documentation.

4609

That was an excellent article - too bad you don’t write that kind of stuff any more - which not by chance is complaining about perhaps the most problematic component of WITE 1. To be very honest with you, I did personally did not understand what that paragraph in the manual you quote meant until I spent quite some time experimenting. You weren’t the only person out there running the same combat 10 times, but what you did not do, which I did, was to repeat the experiment switching off ground support (I am not sure you could do that when you wrote the article, though).

It is an old article - written 7 years ago, barely five months after the release of the game. Today WITE 1 - or rather, WITE 1.5 after Dominik took to himself to keep improving the game - is a game much more capable of explaining itself. And my remark was made in the present, not the past.

For instance, sometime in 2013 it was added the ability to see the “combat details”, which broke down the effect of enemy combat power for you understand the impact of air power by means of uncovering one of the “secrets” of WITE 1. Disruption - a temporary parameter that tracks the degree of cohesion of a combat unit (the higher the disruption, the lower cohesion and generally, ability of the unit elements to participate in battle) was now in plain sight.

Before that the manual did refer to it, but in a scattered manner which did not really help players appreciate how important that factor was (and is) or that it was tracked on a per element category basis. So it could be that your artillery pieces were more disrupted that the infantry due to air attack, leading to less batteries not being ready to make a contribution to the unit firepower as big as it could. Since combat happened in a number of stages, at ever decreasing ranges, disruption conveyed a decreased ability to disrupt attacking elements as they were coming to “contact”. And this in turn had the effect of those attackers doing more damage and disrupting more of your unit elements, and so on…

Setting the air commitments was a rather obtuse way to talk to the AI, and to this day, I don’t know exactly why an overview of how aircraft allocation happened wasn’t included the manual. The manual was being developed by volunteers, and the forums were very active… for later revisions of the manual somebody did the job of combing the forum for nuggets of feedback and clarifications, but it is not a very fun stuff to invest much time on. So now, barring firing an email at Gary Grigsby or Pavel Zagzin, I think you will be out of luck getting a precise answer. But from memory, as I think I read a very long detailed explanation years ago, the most important factor was range to the target hex.

Other aspects of the game were way better documented: there were even examples illustrating the rules :) While I was translating the Ground Combat rules documentation/discussion I added quite a few more examples of rules resolution that I verified personally with the engine. I want to think that some Spanish speaker read those examples and found them useful to understand the range of possible outcomes to rather abstract rules. That took me quite some time!

2by3 I think raised the bar significantly with the documentation for WitW. Not incidentally, you were presented with full page guides that focused on transmitting the essentials to the players on gameplay. Note that in WitW we have a separate air phase too… which probably takes way too long to resolve, but that’s part of the other response :)

To finish a very long post, there’s also this thing with Gary & Joel’s games that I kind of like. You need to experiment to figure out things. Compare playing Minecraft in 2011 to doing so in 2018. Nowadays you go to a wiki and you have an exhaustive list of recipes for crafting stuff. Do you want to achieve X? Easy, go there, and do backwards chaining :) Back in the day, well, you just had to go and find out. I like games that tickle my curiosity… but making those games is a delicate balancing act with royally annoying many other players.

4610

I invite you to tabulate a list equivalent combat results in WITE and OCS - Guderian’s Blitzkrieg, Case Blue, Hube’s Pocket or Baltic Gap. Your choice of game :)

Meanwhile, sorry, but no I’m not buying that statement at face value.

4611

For full disclosure: I was a beta tester of WITW… and I spent a long long time prodding Joel to produce good explanations for combat results. And I am also a beta tester of WITE 2.0.

4612

I think we are talking past another. What’s your point? What’s the mechanical benefit of the simulation depth? Realism?

At the end of the day an attack is going to have certain probabilities of success. That the probabilities are different in different games does not mean you can’t generate a simpler model that replicates WitE results with much less need for detail. A simple model with similar results is obviously possible.

If you are telling me the simplification is not possible and that the difference between 15 and 20 operative tanks is significant enough for the granularity to matter, then the model is most likely flawed (because it is objectively failing to simulate other factors like terrain or troop drunkenness at the same granularity level).

Now, as Bruce has said many times, there’s an immersion factor thanks to that granularity that I really appreciate. But it is most likely irrelevant mechanically.

4613

Hmmm, not my intention to talk past anyone. It was a direct challenge to an statement you made.

I honestly think that is not possible, it is not just that the probabilities of success are different. The range of outcomes and long term effects are different too. The histories generated are incommensurable, as you would not be able to show, formally, a 1 on 1 mapping between the events that make up each history. Which means that the more abstract cannot generate the same same set of artificial histories the computer simulation can. You can argue the gap is immaterial, but you cannot show it.

I suggested you a method to prove me wrong. Good luck.

No, I am not saying you cannot simplify. You can simplify and obtain something different.

Regarding mechanic impacts: I do find the way fatigue, losses, and mechanical tear and wear to have a marked impact in how I play. It less apparent than having combat factors reduced 50% due to a “step loss”, but it has its impact nonetheless.

4614

Because of the randomness inherent in these systems, if you give me the ability to choose dice results in a more abstracted system I can obtain the same results of the more detailed system at the higher level of abstraction. Of course the higher level system can’t show the lower level results, but at the same level the results are indeed mapable by definition.

If there’s a non-mappable outlier, that would imply the existence to too much chaos in the detailed system that would hamper readability of cause and intention anyway.

4615

Actually, the designer is Joseph Miranda and is from his creative period at Strategy & Tactics, (ca no 135 to 200). The game system is really good for operational warfare during the Roman period.One way to sample it is to purchase one of the original Strategy & Tactics games as Ancient Wars is a compilation/expansion of four magazine games:

Trajan - Trajanus war against Persia. This is an excellent introduction to the system. S&T 145
Roman Civil War - Ceasar vs the Senate. S&T 157
Ceasar in Gallia - S&T 165
Germania - The campaigns after the Teutoberger Wald disaster. S&T 175

If you are interested in the topic I think you should visit Boardgamegeek and download the rules for Trajan so you can get rid of the ghost of Berg in you mind.

Here is a good introduction to the game system:

Germania

4616

OK time to reinstall WITE.

4617

The system kept changing and the compilation retrofits some necessary rules. Specially, I think in Trajan by itself you don’t still have the intermediate combat rules that I think do a great job of replicating some difficult to represent outcomes.

4618

We don’t need randomness to show the fundamental limitation I am pointing out. To clarify, the argument is the following:

Given any computer based simulation S1, one can construct a “more abstract” simulation, S2, that can be executed by a human with zero errors guranteed and within human lifespans, that for every sequence of simulations states and inputs x0, u0, x1, u1, …, xn, un generated by S1 there exists another sequence of states and inputs y0,w0, y1, w1, …, yn, wn generated by S2 such that the following four statements hold simultaneously:

i/ |Y| << |X|, |W| << |U| (possible states and inputs have a huge difference in size)
ii/ a feasible function g() that maps pairs xi, ui into pairs yj, wj (this is the easy bit)
iii/ a feasible function h() that maps pairs yj, wj back into pairs xi,ui (note, this does not need to be surjective, you can have several possible maps for a given pair yj, wj).
iv/ one can verify in a reasonable amount of time that h(g(xi, ui)) maps to xi, ui

where feasible function and reasonable time means that one can perform the computations in a number of steps that doesn’t grow crazy as the size of X and U increases.

My claim is that in order to make iii and iv true you need to falsify i (the sets of states and inputs need to have a comparable size) and therefore compromise the assumption that S2 exists (it can be executed manually by a human within humans lifespan and guaranteeing zero errors throughout).

Edit: Credit for the argument goes to the magnificient takedown by Larry Hauser of John Searle’s famous Chinese Box thought experiment

4619

YMMV!

4620

You don’t need iii and iv, because the S1 simulation is not some ontological state you want to check against. What you want to check against is real life historical results. I don’t need to make certain you can go from abstract to concrete, because the whole point is that the granularity is window dressing for the decisions that matter and a more abstract system can get the same results given the right rolls (randomness is super important here, your whole set of assumptions does not take into account randomness and thus posits a more predictable and deterministic outcome).

That we can’t make sure the abstract results can be mapped to the granular system is irrelevant. What’s relevant is that you can get the same results with less effort.

Your point starts with the assumption that a complex simulationist system gives more realistic or better results than an abstracted one. I’m challenging that starting assumption. When the system has imcompatible levels of detail (tracking single men but terrain in 16 km increments) it’s obvious that the simulation will have incredible levels of fudge anyways, more so that a system where scale granularity is consistent between all the represented elements.

A concrete example. Take an operational game where a successful attacking division gets disorganized, representing material and cohesion loss at a level that hampers offensive actions. Take the same attack in WitE. You will have very specific levels of loss and granular need for replacement. But the end result is still a loss of offensive capacity, and the abstract result (general disorganization) is going to be closer to the historical truth of the attack than the granular outcome (where numbers are going to be significantly off from historical outcomes in terms of actual loses). Eventually you’ll have to spend resources/time to refresh the division. The end result of both simulations is he same, but in the process one was more divergent from observed real behavior than the other.

4621

Excuse me, but calling irrelevant what contradicts your claim of equivalence just because isn’t a great argument. Also, it is not very gracious, something of which I take note for future reference in future exchanges.

Just to be clear: my reward for engaging in this discussion is that I am looking forward to learn something from it, not to win any Internet Points. If the discussion is about winning IP, then we’re done, as that stuff does nothing for me.

Where I wrote that ?

Well, physical models of stuff can have lots of “fudge” too and still can serve us well. Newton fudged quite a bit and still you can use his work to make useful predictions and understand billiards. If you are so inclined, do a search on “wildfire dynamics simulation operations insight”, a less mundane matter than parlour games.

In any case, you just “kind of” admitted the validity of my argument, but belittled it calling it “obvious”.

The first problem with this example is that losses as generated by well researched models are eerily close to the historical baselines. That is fact others and me have verified, to our surprise. You made this assertion on the basis of what exactly? Half remembered experiences with games that have changed significantly over the past 7 years?

The second problem is that you are in contradiction when you say that the more abstract outcome of disorganized is “closer to the truth”. The truth is that combat units don’t break down instantly and their combat performance doesn’t always degrade in the same way or according to the schedule allowed by the level of granularity chosen for turns. That variability and diversity in the outcomes have mechanical impacts as the actual gameplay/what happens in the game changes too (the sequences of states and inputs).

You call it texture “without substance”, but I am not sure what is the “substance” you find missing there.

Last thing in your example assumes - seemingly, it may be me misreading you - that players in WITE aren’t constrained to conserve their forces, something that your abstract disorganized unit game state may do by creating a visual signal and rules that detract from further using that unit.

In WITE there is no hard constraint other than a decrease in combat factors: the signal is there, it is just not jumping at you neither the game holds your hand. Most human to human games end not in victory or loss, but by abandonment, as one player eventually realises the he killed the donkey three turns back and further caning serves no purpose. Failure states are subtle in strategy games, more so in war games.

Presenting failure states to players in a clear and timely fashion to avoid frustrating them is pretty much an open problem in game design, imo.

4622

WITE is an interesting case. I should reinstall it and take a poke at it again, as I bounced off of it hard the first time. That was after I even spent a fair amount to get a full-color printout of the manual (dumb, I know, as it changes, but it seemed like a good idea at the time). I was put off among other things by the tracking of individual rifles and stuff like that. I have never been convinced that that sort of detail is necessary for the type of military sim that I want. I know folks love it, and I know as Miguel says that having that sort of deep dive data can, in theory, result in more robust and thorough consideration of variables and lead to strong combat models. In practice, as I perceive it, it often just leads to more stuff. Now, Grigsby and the people working with his stuff are much better at this than, say, Norm Koger, who really is the king of add up how many bayonets and tins of bully beef you have, so I’m willing to try WITE again.

It’s just that I want to play the mind behind the big battle, not necessarily the quartermaster, you know what I mean?

4623

Concur 100%. I even gave it a second chance two months ago, and now it won’t even run well with 1080 screen resolutions. You have to go fiddle with visual settings to make the damn thing run (well, to see everything and operate the GUI). Hard Pass.

90 bucks for a game where you are have to BDE S4 information while commanding Army Groups? And it doesn’t even run well? No thanks.

4624

Intent, game design instead of modeling as much as you can just because. Complex models don’t match actual loses eerily close, the best results of complex data driven military models have a victory prediction accuracy of about 60% to 70% when facing full datasets. Given that baseline random chance would be 50%, military loses and victory modeling leaves a lot to be desired in absence of proper umpiring.

Any model you choose is going to have assumptions built into them, and detail does not really add much when the simulation uses a much higher scale for outcomes anyway. The better the abstraction and the more front loaded those assumptions are the clearer the game design and the exploration of the model than can be realized through playing.

So I’ll just ask. In your opinion what’s the thesis WitE tries to explore regarding to the model it chose? Where is the focus of the design? What are the core assumptions of the model? Because it has to be there, but it’s so buried in the details that it’s obfuscating.

Or basically this:

Obfuscation versus intent. Game design versus modeling. For me there’s no advantage in the lack of restraining other than making the game state less apprehensible. That visual signal for me is a feature, a signal that the designer understands the implication of the model it chose and it’s not leaving it open to interpretation. Now, you have a different opinion, that’s fine. I do understand the appeal (and again, I did like WitE quite a lot, even if I see most of the detail a narrative and played without paying too much attention to it).

4625

I think that’s where this Grigsby Declaration of the Rights of Simulations ends. It really is all down to preference. There is no “objectively superior game or method of design”. This discussion really hit What I like-land 3 days ago, @MiquelRamirez :)