The Sheriff of Oosterbeek – A Scenario Design DAR/AAR

[Pak40]

Also, for the Allied side, it doesn't matter if the artillery assets firing are on-map or off-map. The Allied AI will always seem to pick the same plan when the weight is the same.

perhaps a bug? Need more testing and a quick alert to BF so they can fix it in time for MG

[JonS]

17 – Testing

"The reason the American Army does so well in wartime, is that war is chaos, and the American Army practices chaos on a daily basis."
from a post-war debriefing of a German General

Before getting into testing, I want to air some thoughts on CM as a game vs CM as a simulation. WWII was vast and complex, and some really unexpected things happened from time to time. But overall, there were clear trends about what worked on the battlefield and what didn’t. Just because something actually happened in WWII does NOT mean it should happen every time a scenario of that event is played. This is one of the issues with creating strictly Historical scenarios.

Consider the battle when Dick Winters overran that battery behind UTAH, or when the British paras overran the Merville Battery. Both of these were outstanding actions. But they're outstanding actions, and remembered as such, because they shouldn't have been won by the Allies. In either of those actions slightly different decisions by any of the actors on either side could have resulted in an easy German victory.

In CM terms, those 'slightly different decisions' are the command inputs the player uses, and the way the TacAI reacts to those prompts. A decent recreation of those battles as CM scenarios should - I think - be really hard for the Allied player to win. At the very least it should take more than just turning up with your uber 'leet paras and pressing 'Go'.

CM is a complex game, and it takes skill and effort to learn to play it well. As noted elsewhere, this means that some players will take a while before they can win difficult scenarios. Some players may never be able to win some scenarios, because they never gain the skills to combine orders in ways that create the effect they need. And that’s ok.

It's not about 'solving the puzzle', although sometimes it might seem that way. To play CM well you have to give precise, accurate, timely and relevant orders to your pixeltruppen, and choreograph your available resources to create a unified team that is working together, rather than a collection of individuals. One of my regular opponents is a wizard with armoured vehicles, and I usually struggle to create the effect I want – dead German tanks – because he is much better at creating the effect he wants – dead Allied tanks. Sometimes the difference between winning and losing might come down to the difference between using MOVE or HUNT, or giving a particular order on turn 10 instead of turn 11 ... or turn 9. I think that kind of chaos and unpredictability is part of what makes CM such a compelling game.

Winning a scenario based on Merville Battery, or Winter's attack on the German battery, should be a possible outcome, but it shouldn’t be the expected outcome.

The point here, as it applies to testing, is that I’m not looking for a particular outcome each time I test. Instead I’m looking for coherent play that leads to a plausible outcome.

Right, on with the testing. Although testing is described here as a discrete activity, in practice testing of one sort or another is going on pretty much constantly alongside all the other scenario design activities.

There’s a number of simple and quick tests that are useful to make sure you’ve gotten the basics right.

The first explicit test for this scenario was to see how long it took to walk (Move) from one end of the map to the other along the Benedendorpsweg: 42 minutes, or about half the length of the scenario. I’m reasonably happy with that. The German player won’t be able to muck around, but with skill and some luck they should be able to get in amongst the British gunline. Also, the trucks and halftracks provide the German player enough seats for 55 passengers in a single lift. That’s enough to shuttle a platoon(+) at a time, as long as the transport is looked after. However, some of the German infantry doesn’t arrive until the 20 minute mark, and the final elements of KG Harder don’t turn up until the 30 minute mark, which is pushing the timing envelope even with the transport available. Therefore I added another 10 minutes to the run-time of the scenario, taking it out to 1 hour and 40 minutes.

Another useful test is to ceasefire at the end of the first turn, and see who wins. Usually this works out ok, but occasionally I’ve had cases where the “best” ploy against the AI is to immediately http://www.youtube.com/watch?v=NHWjlCaIrQo. That seems to defeat the whole point of playing the game, but it’s usually fairly easy to fix if needs be.

Related to that is testing the distribution and value of objectives. For this scenario it’s meant reallocating points significantly so that 50% of the value for terrain objectives are gained once player reaches – or holds – the area designated High Ground. This was the location where Thompson, and later Lonsdale, held the Germans on the 19th and 20th. The thinking here is that the British holding the High Ground is a victory for them (as it was historically), and holding further to the east is more of a victory, while being pushed back to the west is a defeat.

A different, but still quick and easy, test was to see if vehicles could climb up on to the top of the rail embankment. For various reasons I don’t want any vehicles to be able to get up there. Starting the scenario as Germans, then placing a waypoint at various locations up on top of the embankment showed that the StuGs could actually get up the top, so a thin diagonal strip of Heavy Forest (without any brush, shrubs, or trees) was added to the ‘British’ side of the embankment to stop that.

17.1: LtCol Thompson exhorts the defeated paratroopers to turn around and dig in.

The main and most time consuming component of testing is seeing how the scenario actually plays. I break this down into two activities – Static Testing against the AI, and dynamic testing against the AI or preferably against another human.

Static testing against the AI utilises the ‘Scenario Author Test Mode’ skill setting to watch what the AI does, when it does it, and how it goes about doing it. It’s a pretty simple process. I start the battle as I the British player, in Scenario Author Test Mode, using WEGO because it’s much faster than Real Time.

17.2: Starting the battle in Scenario Author Test Mode. This functionality is a godsend for scenario designers.

The key point here is not to do anything with the British forces. They’re just left in their start positions. I’m not trying to prove to myself that I can defeat my own AI plan. That’d be trivial.

Instead the Scenario Author Test Mode is used to observe how the AI-controlled side manœuvres, and how it deals with the enemy forces it finds. This can be run through pretty quickly too – I’m not too interested in the second-by-second, bullet-by-bullet ebb and flow. In fact getting down in the weeds can be confusing. Rather I’m watching the minute-by-minute development of the overall battle, and how the AI advances its forces to apply combat power. Because of this it’s often not necessary to watch the battle at all – fast stepping in 5-second time chunks, or mashing the red button to quickly cycle through the turns gives a pretty good idea of whether the AI plan coordinates the forces available, or if it still needs some work.

Predictably, this throws up a lot of outright gaffs, and areas for improvement, especially in the first couple of trials.

17.3: Various issues that came to light during the first three test cycles of the first 20-30 minutes of German AI plan No.1. Issues are noted, but the scenario is allowed to continue to try and catch as many issues as possible each go-around.

This soon becomes a cyclic process of adjusting the AI, then running the battle to see how it goes and taking notes on any issues, then adjusting the AI again, running the battle again, and so on. Hopefully the number of issues raised each go-around reduces. I’m not looking for perfection from the AI – that’s impossible. An active human opponent could be doing almost anything, so the best that can be done is to create a plan which sees the AI controlled forces securing a series of objectives in a reasonable manner that should be sufficient to overcome reasonable resistance. You also need to be realistic about what can be achieved – testing is incredibly time consuming, and there’s only so much that can be done with a strictly linear script. Determined players will always find ways to ‘break’ the scenario, to beat it trivially by doing something unexpected or impractical against an active opponent … but don’t worry about that.

So, that’s all good, and the German AI plan progressed from utterly inept to reasonably competent. Most of the tweaks and adjustments were to do with making the Germans more competent when compared to their opponents (mainly by decreasing the experience or motivation of the British, and the amount of ammo certain units are carrying), or producing better co-ordination of the different AI teams in time and space. Once the first phase of the first plan has been tested, then it’s on to the second phase, then the second plan, and then the Allied AI plans.

While static testing the AI, also try and enroll a couple of people to concurrently test the scenario H2H. Playing through a scenario can take quite a while, so it’s a good idea to get this underway as soon as practical. It’s a good idea to make friends with a few tolerant and reliable players who have an eye for detail and are able to present useful feedback. It’s also a good idea to give guidelines on what sort of feedback is required. Examples of some of the kinds of feedback I find useful are:
. 1- Is the briefing clear and helpful? Eg. Do you know by reading it exactly what you need to do? Do you get enough info about forces and reinforcements?
. 2- If you ask for a cease fire right at the beginning, what result do you get?
. 3- Are the forces balanced? Is it reasonable to expect that players will find the mission easy/medium/hard?
. 4- Are any reinforcements arriving on time?
. 5- Is the time given enough to achieve the objectives?
. 6- Is the AAR at the end of the mission ok?
. 7- Are the points assigned to objectives "fair", reflecting the goals of the mission as stated in the briefing?
. 8- is the overall scen consistent with the intent of the mission as described in the first post of this thread? Are the forces on both sides as suggested, is the setting, date, and time correct?
. 9- is the spelling of reinforcements, landmarks, objectives, unit names, etc consistent and correct?

Feedback is always going to be subjective, but it’s always useful even if you chose not to act on it. I had a couple of people testing this scenario, and they’ve been invited to chime in here with their observations on the role of the scenario tester in general, and things they found while looking at this scenario in particular.

After a few run-throughs of the German AI-plan, static testing threw up a rather significant issue. After watching the AI’s efforts on this large map, and assuming a moderately competent opponent, I don’t think it is reasonable to suppose that the Germans will ever get much past about midway, even though technically there is time for them to walk all the way to the Church and back again.

There are also a lot of forces in play, particularly the British. The Air Landing Light Battery, in particular, consists of seven guns and associated ammo teams, about ten HQ teams, and over 20 jeeps. All up there’s getting on for 50 units just in the gun battery. The combination of unit quantity and map size is causing some performance issues.

17.4: 2nd Air-Landing Light Battery deployed in all its glory. There should be another 50-odd Glider Pilots, represented by roughly 20 more CM-units, located in this vicinity helping with local defence of the guns. Ideally, none of these units will play any part in the battle. A complete waste of time.

Furthermore, the guns shouldn’t actually do anything in the scenario. They’ll just sit there, and maybe get involved in the fight if the German player does really well. Or a canny British player will use them in a wildly a-historical fashion by moving them to flesh out the forward defence lines.

All of this is a problem, and it’s caused me to rethink my concept and what I’m actually trying to depict here. I’ve realised that I became too fixated on Thompson as a artillery commander, and lost sight of his role as a leader. The real battle in lower Oosterbeek on the 19th and 20th was to establish a coherent defence line to protect the divisional enclave, and for various reasons that line happened to end up forward of Thompson’s guns. But the guns themselves weren’t, strictly, part of that battle.

So, as a direct result of this testing I’ve brought out the chainsaw to dramatically reduce the size of the map, and with it the size of the British forces. The map has been cut down on the western and southern sides, from 1808 x 1040m to a new size of 1264 x 640m, meaning the reduced map is just 43% the size of the old map. As much as it pains me, the on-map Air Landing Light Battery has been removed and replaced with an off-map module of 4 guns, one of the Glider Pilot Squadrons has been deleted, and various support weapons are also gone. In addition the length of the scenario has been reduced from 1hr 40 mins back down to 1hr 10 mins, and may come down some more.

With the western third of the map (including the Church and the petrol station) gone the terrain objectives have had to be substantially re-worked, but fortunately almost everything else is still usable as-is, or with only slight adjustment. In particular, the German AI plan I was working on is still fully functional. The cycle of testing, tweaking, testing, adjusting continues with that plan until it’s complete, while incorporating feedback from my H2H testers, and then testing additional German AI plans and the British AI plans.

17.5: The reduced map, showing locations of the new British objectives, which now more clearly focuses on establishing a viable defence line.

Doubts that the larger version would ever really work as either an H2H or vs-AI scenario had been growing during development, and testing confirmed those doubts. The new smaller and tighter scenario ‘feels’ a lot better, and testing proceeds once more. In the short term all my design effort is going into the small one, although the large version may be finished and released eventually.

I’m glad this was picked up fairly early in testing, but I do wish I’d spent more time thinking about this at the outset – with over half the map now gone, a LOT of time could have been saved.

Note to self: think harder about your concept next time.

Back to start of thread Edited January 8, 2015 by JonS

[Sublime]

Jon I have to admit I simply enjoy reading your posts - I actually have no desire to create scenarios, I just like how you put your posts together. Very well written. Nice quotes, etc.

[Childress]

Me neither. Jon's account may prove more discouraging than incentivising for would be designers. What a detail-sweating and time consuming endeavor! You need to understand the game inside and out. A good read, tho.

[Rocky Balboa]

This thread absolutely underscores the value in good scenario design. Players sometimes focus too much on game visuals, artwork and program modifications. We also sometimes fail to realize how much work, time and dedication goes into making well crafted and researched scenarios and campaigns.

Thanks, Jon S for the awesome dev diary and I can't wait to play this scenario ....

[kohlenklau]

That is your hand-writing? I didn't know your area was in a seismically unstable region. Or perhaps a nip of the ol' sauce, eh headmaster?

Thanks so much for this extreme effort to produce a road map for the lost souls of scenario city.

[Ithikial_AU]

17 – Testing

Before getting into testing, I want to air some thoughts on CM as a game vs CM as a simulation. WWII was vast and complex, and some really unexpected things happened from time to time. But overall, there were clear trends about what worked on the battlefield and what didn’t. Just because something actually happened in WWII does NOT mean it should happen every time a scenario of that event is played. This is one of the issues with creating strictly Historical scenarios.

So in short don't aim for the scripted Hollywood retelling of events?

Shame you've had to slice and dice part of this scenario. When constructing scenarios as part of the official release design group are their hard limits set in place on things like size, to account for consumers hardware and advertised system requirements of the product? When testing larger scenarios would you try to aim to get testers with different PC specs to see how much of a performance hit consumers are likely to get on a mid or low end system?

Still loving this series and would be a great PDF manual when it's done! Keep up the great work.

[JonS]

So in short don't aim for the scripted Hollywood retelling of events?

Basically; yes. I think in one of the earliest posts I talked about setting the stage, rather than writing the script? With the research I'm interested in what units got to the battle, and how they got to be there, rather than what they did once they arrived.

Shame you've had to slice and dice part of this scenario.

In some ways, yeah. But it's been a nagging concern for a while that the scenario wasn't really adding up. The scenario as it is now is really quite in tune with my original intentions, and the play seems to be quite good, so I'm really happy with that. I'm a bit sad about cutting the guns, and large chunks of the map, but they were always means to an end, not the end itself.

When constructing scenarios as part of the official release design group are their hard limits set in place on things like size, to account for consumers hardware and advertised system requirements of the product? When testing larger scenarios would you try to aim to get testers with different PC specs to see how much of a performance hit consumers are likely to get on a mid or low end system?

No, and not really. People design the kinds of scenarios they're interested in, although they usually try to extend themselves to try different things, sizes, units, and whatnot, all without official prompting. As for testing on different rigs .. apart from being an NP-complete problem, experience has shown that it isn't really necessary. There are loads of options to dial down the graphic quality on slower rigs to make large scenarios playable, the very definition of 'playable' is highly variable and personal, and there's generally a pretty good range of different sized scenarios anyway.

Still loving this series and would be a great PDF manual when it's done! Keep up the great work.

Thanks! There is a plan to bundle it together as a PDF, but that'll take some time to re-write the tense, modify the UBB codes, and if I get a chance to make some of the imagery a bit slicker.

[JonS]

That is your hand-writing? I didn't know your area was in a seismically unstable region.

Ha. Well, the building I'm working in has some rather alarming new cracks courtesy of the earthquake we had a week or two ago ... but that has nothing to do with my abysmal handwriting!

[JonS]

stuff

Thank you for reading it!

[JonS]

Jon's account may prove more discouraging than incentivising for would be designers.

Mission NOT accomplished

You need to understand the game inside and out.

That helps, sure, and this thread is at least partly intended as a short cut so other designers can leverage off what I've figured out and what I've picked up from other designers, instead of having to figure it all out from scratch again. But actually you don't really need to know the game inside out.

A good, clear idea coupled with modest scope (i.e., a smallish map and ~ company sized forces) will carry you a long, long way. The rest is really just patience and an inquiring mind.

[para]

I think it is brilliant what JonS has done. I have to admit though, it has put me off ever trying my hand at scenario designing. It is far too hard.

[ASL Veteran]

I think it is brilliant what JonS has done. I have to admit though, it has put me off ever trying my hand at scenario designing. It is far too hard.

Jon has done an excellent job of organizing the process he uses when making a scenario, but while most designers may go through each of the steps he has listed it is more typically done in a less methodical manner. I think probably the most difficult part isn't following the steps per se, but rather acquiring the ability to 'feel' whether what you are doing is going to end up with something with potential or not. Sometimes you are plugging away and it just feels right from beginning to end and everything goes smoothly. Other times you sit there and struggle with something chipping away at this or modifying that until it either finally feels right or you have to abandon it. If you follow the steps Jon has outlined your chances of getting it right from the beginning are much better than if you just open up the editor and start flailing away. However, after you have done it for a while it should become more of an intuitive process than a mechanical one.

[PanzerMike]

JonS has inspired me to create my first scenario. It is just finished, but needs playtesting. And tell you what, until 5 minutes ago I never even knew there was a special play mode for scenario testing!

[Childress]

This is a very ambitious scenario, and Jon has paid meticulous attention to accuracy, terrain, actual formations, etc. He's also a certified grog. Maybe his ouevre-in-progress will, if discouraging to rank beginners, stimulate designers who already have some some modest scenarios to their credit. Especially if they're retired on a generous pension or trust fund.

[agusto]

@PanzerMike: Yeah this thread was very inspirational and helpful. As a result of it I ve finished one scenario during the past month and another one is in production and nearing release. You can check them out here (incredibly shameless & thread hijacking self-bump follows):

http://www.battlefront.com/community/showthread.php?t=110566

http://www.battlefront.com/community/showthread.php?t=111001

@ASL_Veteran: Using a methodical approach when designing scenarios IMO greatly increases work effectiveness, i.e. it directly reduces the amout of hours you have to spend making it. That is basically why this thread motivated me to start making scenarios again. My first unmethodical try a couple of months ago was quite unsatisfying and very slow, basically because i had no real idea of how to do it and i just starteted with no real concept or anything but just a very vague idea. It is always hard to do something when you dont have a real idea of what you actually want to do .

[snake_eye]

@JONS

Catching up with your very interesting post, I have a question, being surprise that you found WEGO faster for testing than RT while using Scenario Author Test Mode. I would have thought the contrary. I prefer RT being able to watch the battle unrolling. Wego on the contrary allows me to look at an area, I might have overlooked if there are movements in different areas.

Am I wrong ?

Quote "Static testing against the AI utilises the ‘Scenario Author Test Mode’ skill setting to watch what the AI does, when it does it, and how it goes about doing it. It’s a pretty simple process. I start the battle as I the British player, in Scenario Author Test Mode, using WEGO because it’s much faster than Real Time."

Cheer

[JonS]

Snake_Eye,

RT goes as fast as it goes. But with WEGO I can fast forward, or just turn-mash (press the red button as soon as possible). I'm not to worried about how the turn plays out, because there is no way I can know where a Human player will have his units on any given turn.

I'm just checking to see that units are moving from point to point coherently and in a reasonably tactically sound way, and that where necessary they're moving in coordination with other AI groups.

With turn mashing I can smash through a 'game' fairly quickly, taking notes of issues as they arise.

[JonS]

18 - Polishing

”I don't write a quick draft and then revise; instead, I work slowly page by page, revising and polishing.”
Dean Koontz

With AI plans complete, and testing done, the scenario is about ready for release. But because the little things matter, it’s worth taking some time to check the scenario over to make sure that it is nicely polished. It’s highly tempting to rush this and just release the scenario, but it’s really annoying to have to upload a second iteration of anything to the Repository because you miss-sbelled somefink. Unlike Mr Koontz, I tend to draft and revise, working fairly quickly to get a scenario that’s fully functional, but which needs a lot of polishing before it’s ready to step out on the stage.

Proofread everything. The briefings, obviously, but also the briefing imagery, the on-map landmarks, reinforcement arrivial notifications, objectives, unit names, everything that has been typed in. Spelling and grammar are obvious concerns here, obviously, but also check the use of any foreign terms

Consistency. Check that the spelling of names is consistent in any imagery used, on map landmarks or objectives, in the briefings, in the units editor, and so on. Also, check that a consistent story is being told in the briefings, in the objectives, and on the map.

18.1: The name and spelling of 1 Kp. (Pz Crews) is consistent in the Briefing, in the Unit Editor, in the reinforcement slot, and in the Unit Objectives editor.

Coherency. Take a last chance to check that the ‘story’ of the scenario makes sense. All the various elements of the scenario should work together to tell a coherent story that is plausible and engaging. This is a judgement call, but listen to your gut.

Check that all the total pool of points adds up to exactly 1,000 for each side, and that overall the weightings still look right – that they reward success, that they tell the story you want told, and that that story will be consistent with the scenario briefing. In this case, because the map and forces have changed so much, this is something that had to be paid particular attention, and was in fact badly wrong when I checked it. I also needed to modify the location of some of the landmarks because they were either now off map, no longer central, or clashed with the new terrain objective locations.

18.2: Overview of the final, reduced, map for The Sheriff of Oosterbeek.

And just like that – the scenario is finished!

Back to start of thread Edited January 8, 2015 by JonS

[Ithikial_AU]

Looks great JonS. So how long until release?

Thanks for all your hard work with this series.

[Holien]

Are we there yet, are we there yet...

Looks good and I have avoided reading certain sections as I want to really enjoy this scenario and then come back and understand how it has been constructed...

Thxs to you and all the other designers who feed my addiction...

[altipueri]

I am impressed at the amount of work that goes into this scenario making - well done.

[Pak40]

Thanks! There is a plan to bundle it together as a PDF, but that'll take some time to re-write the tense, modify the UBB codes, and if I get a chance to make some of the imagery a bit slicker.

I've already used Print to PDF to to convert most of the posts. The imagery comes in fine. It's not too hard to do but I still need export the first two and 15-18.

[sburke]

This is a very ambitious scenario, and Jon has paid meticulous attention to accuracy, terrain, actual formations, etc. He's also a certified grog. Maybe his ouevre-in-progress will, if discouraging to rank beginners, stimulate designers who already have some some modest scenarios to their credit. Especially if they're retired on a generous pension or trust fund.

Well to repeat something I mentioned earlier that Jon had taught me - start small. Your first scenario should not be to attempt something like Oosterbeek. A platoon scale battle would be much more feasible and a good place to learn the basics. The process is pretty much similar, but with a lot less overhead at each phase.

[sburke]

Feedback is always going to be subjective, but it’s always useful even if you chose not to act on it. I had a couple of people testing this scenario, and they’ve been invited to chime in here with their observations on the role of the scenario tester in general, and things they found while looking at this scenario in particular.

After a few run-throughs of the German AI-plan, static testing threw up a rather significant issue. After watching the AI’s efforts on this large map, and assuming a moderately competent opponent, I don’t think it is reasonable to suppose that the Germans will ever get much past about midway, even though technically there is time for them to walk all the way to the Church and back again.

There are also a lot of forces in play, particularly the British. The Air Landing Light Battery, in particular, consists of seven guns and associated ammo teams, about ten HQ teams, and over 20 jeeps. All up there’s getting on for 50 units just in the gun battery. The combination of unit quantity and map size is causing some performance issues.

Furthermore, the guns shouldn’t actually do anything in the scenario. They’ll just sit there, and maybe get involved in the fight if the German player does really well. Or a canny British player will use them in a wildly a-historical fashion by moving them to flesh out the forward defence lines.

All of this is a problem, and it’s caused me to rethink my concept and what I’m actually trying to depict here. I’ve realised that I became too fixated on Thompson as a artillery commander, and lost sight of his role as a leader. The real battle in lower Oosterbeek on the 19th and 20th was to establish a coherent defence line to protect the divisional enclave, and for various reasons that line happened to end up forward of Thompson’s guns. But the guns themselves weren’t, strictly, part of that battle.

So, as a direct result of this testing I’ve brought out the chainsaw to dramatically reduce the size of the map, and with it the size of the British forces. The map has been cut down on the western and southern sides, from 1808 x 1040m to a new size of 1264 x 640m, meaning the reduced map is just 43% the size of the old map. As much as it pains me, the on-map Air Landing Light Battery has been removed and replaced with an off-map module of 4 guns, one of the Glider Pilot Squadrons has been deleted, and various support weapons are also gone. In addition the length of the scenario has been reduced from 1hr 40 mins back down to 1hr 10 mins, and may come down some more.

With the western third of the map (including the Church and the petrol station) gone the terrain objectives have had to be substantially re-worked, but fortunately almost everything else is still usable as-is, or with only slight adjustment. In particular, the German AI plan I was working on is still fully functional. The cycle of testing, tweaking, testing, adjusting continues with that plan until it’s complete, while incorporating feedback from my H2H testers, and then testing additional German AI plans and the British AI plans.

I was one of those testing this scenario. Our initial feedback I think did help confirm a couple things for Jon and some of my feedback was specific to the allied side as that had been my role. One of the things I tried to focus on was the intent of the scenario. In this regard Jon's briefing was really important as it laid out the situation and what issues I would need to deal with.

Basically my situation initially was supposed to be complete chaos. I needed to start organizing my defense and while doing that I needed to grapple with the ammo shortages I was supposed to be facing. What I found was my ammo situation was not as critical as it should have been partly due to the presence of the artillery units. Their vehicles carried some 3000 rounds of 9mm which is exactly what I needed for my sten guns.

The second thing we found was the British defensive position combined with the vast number of automatic weapons I had meant the Germans faced an extremely difficult task. In short we knew pretty quickly the map was too large and the "stationary" forces provided the allies with far too much in the way of resources in HTH play to create the conditions the British should have been facing.

I have been traveling for about a week and haven't seen the new version. I expect the Germans still face an uphill battle, but considering the quality of the British force this is as it should be. The Germans were throwing ad hoc forces against a very highly trained and motivated force in dense terrain.

I honestly don't know how much we contributed to Jon's design he hadn't already figured out, but we at least confirmed some of his thoughts and also showed some of what a human would do that could totally distort the battle using forces not intended to be part of the fight. Personally I initially wanted the gun line as it gave a real gut check to know that if the Germans were to reach that point it would mean the collapse of the 1st AD defenses. However once it became clear the battle was never going to go that far and those resources were going to be used regardless of Jon's intent, they would need to be removed.

And that is another thing I have learned from a number of other designers and testers. If anything is in there that isn't contributing to the battle and is just extraneous, delete it. As an example in working on the Venafro scenario for GL the map actually started outside the town and the initial part of of the scenario was taken up just trying to enter the urban portion. As was pointed out to me, the urban portion was really what the scenario was. Net result, the map was cut down and edited to focus on what was important. Whether anyone is actually enjoying it is a different question, but I learned a lot about AI plans while doing it despite it being a tiny scenario. Cutting down map size seems to be a recurring phenomenon.