Release of the source code, this time with a poll [Archive]

Steph

Apr 02, 2008, 01:56 AM

Now that we are the 2nd of April, time for a little update :D

It was partly an April's fool, that's what made it interesting.

The true part:
- I've had contact with someone from 2k. It's a good news in itself... But the first answer was a bit disappointing. In fact, I had the feeling he did not really understood my proposition, and was focus on "create a company, come with 1 millions $ (he never quoted a figure, that was just my impression), and we can start to discuss how you can buy the licensing to do a new game".
- So I sent a second message, explaining better that we don't want a new game, but a "correction" or a "small" expansion of the existing one.
- Initially I had no answer, but as it was when our email server was not working very well, I send again a message, and I got an automatic answer "out for paternity leave".
- So a few weeks later, I sent it again, asking clearly if he wants to discuss it further, or if it's a definitive no, and then no need to waste our time.
- I was told that he wanted to read my proposition carefully in details, but that he was very busy between the baby and Civ Revolution, and he had not been able to really take time to it properly.

The April's fool part :
- The gathering of 100 people in one day. There was no such requirement.

But I thought that it would be intersting to post the message the 1st of April, to :
- See who would answer directly without realizing it's the 1st (first level of April's fool)
- See who would notice it was the 1st and claim April's fool, when it was for a good part true (second level of April's fool)
- Take the opportunity to inform you a bit of the status of the "discussion" (not very advanced, but at least it's not completly stopped).
- And see what kind of support we have.

I think an important part is to see that many people would buy a fan made expansion, and very few are demanding a free one. That's positive, as it can show that there could be a market and some chance of profit with such an expansion for Take2.
They would not consider something complelty free.

Oh, and for the moment, as long as CivIII is selling, there is NO chance of a public release of the source code. And after it's released, I think the chance are still limited. But I still see hope for a limited release for a fan made expansion.

Quintillus

Apr 02, 2008, 02:16 AM

How much of a chance is there of the source code being updated with a fan made expansion? Such as removal of limitations, elimination of bugs, etc.? I managed to make Civ crash to desktop on a program error twice in the past six hours, so I know there's still bugs in there :D! Probably ought to post those errors now that I think of it.

I'm in it for the long haul at any rate. Probably still will be playing Civ3 in four years, quite possibly longer depending on what Civ5 is like and if I'll be a bit below the recommended system specs for Civ5 like I have been for the last two iterations.

Steph

Apr 02, 2008, 02:23 AM

How much of a chance is there of the source code being updated with a fan made expansion? Such as removal of limitations, elimination of bugs, etc.?
That's basically the essence of the proposition I sent. Correct some bugs, remove some limitations, add or modify a few features...
Nothing really big (at least to start with!) so we won't implement civics or unit promotions à la CIV, but for instance we could add a "can bombard land" "can bombard sea" in the flags, so a submarine can torpedo a carrier, but not a armored division...

Mirc

Apr 02, 2008, 02:24 AM

- See who would notice it was the 1st and claim April's fool, when it was for a good part true (second level of April's fool)

:lol: So no matter what we choose, we are fooled :p That was a good one.

Oh and I'm ready to pay for any of the possible options.

Bjornlo

Apr 02, 2008, 04:07 AM

Well, we'll have to see if this goes anywhere. I strongly suspect it will go no where unless/until we figure out a way for a way for this to be profitable for them. If we can do that, then this might (might) work.

What I thought might work for them was this:
They do a new version of Civ3 with some modified component, and release partial source code /SDK with that. This partial source code allows modification to much of the game, but not the "new" bit. So those who want to play with the new user created content must buy this "upgrade". While at the same time, it remains backwardly compatible with the old content. And might in fact require that some or all of the orignal be there. This would sell Civ3 games and this new Civ3.1 resource kits.
If it would sell enough to be worth it to them, I can not answer. The other option would be that they get the right to re-sell any and all content and changes made with this kit with only nominal/token payments to the creator?
I am not sure about this last bit. I know they have sold user created content in the past. I also know that as far as user modifcations they have been fairly decent about our efforts. They provide more support for the modder than most.

I remain hopeful, but not optomistic.

Steph

Apr 02, 2008, 04:31 AM

Well, we'll have to see if this goes anywhere. I strongly suspect it will go no where unless/until we figure out a way for a way for this to be profitable for them. If we can do that, then this might (might) work.
Rest assure that profitabilty for them is part of the arguments I'm presenting them.

What I thought might work for them was this:
They do a new version of Civ3 with some modified component, and release partial source code /SDK with that .
Not going to happen. Civ3 is not made in a way to allow a SDK, so this would require heavy change, and more resources on their side that they are ready to commit. In fact, they are ready to commit between 0 and zero resources.
That's part of the trick. Originally, they were "stuck" with a business model in mind that was not possible. Hence the initial "no".
I'm trying to present alternate business models that could be beneficial for everyone while respecting their constraints, including the virtually no resource available.

wolf_66

Apr 02, 2008, 04:38 AM

I strongly suspected April's fool and I did not reply yesterday...

Now, apart from jokes, let me say once more I would be glad to pay for CIV3 expanded version and/or any singnificant improvement in game features.
Many civ fans posted similar messages here and in old thread. This should mean something for 2k guys (about possible profits side) but, as Bjornlo wrote: "I remain hopeful, but not optimistic"

timerover51

Apr 02, 2008, 08:22 AM

Ultramarine

Apr 02, 2008, 10:04 AM

:goodjob: I support the idea. If they were able to provide documentation on how to unlock and use some of the functions and features that may have been planed, but didn't make it to the final version, I'd pay for that "booklet".

Ozymandias

Apr 02, 2008, 10:05 AM

Steph

Apr 02, 2008, 10:09 AM

As I've said in the past, I still believe that the best solution (assuming they followed contemporary coding standards!) would be to get a map (if it exists) of the main code's APIs. Realistically, we're not going to muck with source code (reverse engineering algorithms; gnashing of teeth; tearing of hair; etc.). The APIs would at least potentially let us make the editor a more robust beast.

Best,

Oz
It wouldn't help much. We need to be able to go inside the functions to correct them. So yes, we want to muck with source code. I doubt they have API, the code was not designed to be exandable with a SDK like in CIV.

Ozymandias

Apr 02, 2008, 10:16 AM

It wouldn't help much. We need to be able to go inside the functions to correct them. So yes, we want to muck with source code. I doubt they have API, the code was not designed to be exandable with a SDK like in CIV.

Okay, but then this will require (1) a dedicated team of 4-6 programmers (I'm guestimating) (2) a REALLY dedicated project manager and (3) a forum for sanely and fairly deciding what to change and (4) (again a guestimate) perhaps a calendar year of effort.

Best,

Oz

Quintillus

Apr 02, 2008, 10:58 AM

I voted regardless of the April Fool's idea. Based on what I have seen on the board's and downloaded, a fan-designed update is not only feasible, but would probably be very good. I, for one, would be more than willing to offer my military history and weapon effects expertise to the project. As for the source code, maybe we could get rid of the corruption idea or at least make it a little easier to deal with. Even without the source code, a fan-design update is a good idea.

Corruption can already be signficantly reduced/eliminated in the editor. You can compensate with higher maintenance costs if you wish. It's not a perfect solution, as it makes Communism kind of useless, but it's already fairly possible.

I wouldn't say a source code release is necessary, but I do think some modifications therein are necessary. Without any change in the engine, a fan-designed update is limited to what can already be done. Which is amazing, but most of it is already freely available, much of it at CFC and much of the rest at Apolyton. Some updates to the source code, even if it wasn't realeased, could greatly expand the game far beyond what is possible without modifications.

I certainly wouldn't be opposed to bundling some fan-made efforts with an expansion, as that would make them far more widespread than just those who happen upon them on the Web. But it really needs to be more than that.

Okay, but then this will require (1) a dedicated team of 4-6 programmers (I'm guestimating) (2) a REALLY dedicated project manager and (3) a forum for sanely and fairly deciding what to change and (4) (again a guestimate) perhaps a calendar year of effort.

Best,

Oz

Probably true, but a calendar year isn't too much. There was some guy on here awhile back, before I joined I think, French I think, who tried to remove the 512 city limit by messing around with memory. He somewhat succeeded, but there were lots of bugs and in the end he gave up on debugging it and removed the city-limit overcoming part of his mod because it was impossible to keep track of all the references to the 512-city array in memory. It should be considerably easier working with the actual source code than in memory, but it probably still will be difficult. And then there's the matter of trying optimizations of the code. And we don't even know what language the source code is in - do we? - so it's hard to assemble a team yet.

edit: There it is! Civ3Ext (http://forums.civfanatics.com/showthread.php?s=&threadid=57158) by Muchembled. And here (http://forums.civfanatics.com/showthread.php?p=1524534#post1524534) is the post where he discusses his attempt at removing the city limit in memory space. Post #72 also has some info. Unfortunately he's disappeared and the posted version doesn't have the city-limit removed attempt in it, so I'm not sure where one could get it.

vingrjoe

Apr 02, 2008, 11:13 AM

but for instance we could add a "can bombard land" "can bombard sea" in the flags, so a submarine can torpedo a carrier, but not a armored division...

Man how I'd love to be able to have that option in the editor.

That and allowing allied units share the same tile..or at least subs and allied ships sharing the same tile.

Blue Monkey

Apr 02, 2008, 12:03 PM

Ozymandias

Apr 02, 2008, 03:04 PM

RedwallFortress

Apr 02, 2008, 03:16 PM

The real Question in my Mind, is why The Hell do all the people in the CFC Community pay lip service to Firaxis in the first Place?

We have nearly 150,000 Members, Take-Two Interactive has only Employees 2,002, and Firaxis Games only has about 85.

More over, the CFC community is by far more productive.
Take a look at Wyrmshadow, over 1,400 Units made, and still hard at work!!

It is impossible that among the 150,000 members, there aren’t people who know how to Program.
Why then don’t we simply make our own Civ?

I am sure that Programming a game engine like the one in Civ is possible, and we already have a bigger Graphic Library then any thing Firaxis has.

I repeat: Why don’t we simply make our own Civ?

Bungus

Apr 02, 2008, 03:25 PM

I can't program, but I'll make pretty good blitzes for anyone who does.

Stormrage

Apr 02, 2008, 03:52 PM

Blue Monkey

Apr 02, 2008, 04:24 PM

Phlegmak

Apr 02, 2008, 04:29 PM

Chris85

Apr 02, 2008, 04:31 PM

I estimate only about 10,000 of those 150,000 people are still active.

It would be cool if we made our own game, but Stormy pretty much covered why such an effort would be difficult. There's always hope for the thing Steph and those guys are working on.

Steph

Apr 02, 2008, 04:34 PM

Okay, but then this will require (1) a dedicated team of 4-6 programmers (I'm guestimating) (2) a REALLY dedicated project manager and (3) a forum for sanely and fairly deciding what to change and (4) (again a guestimate) perhaps a calendar year of effort.

I don't think we really need a lot of programmers for relatively small changes, but we may need quite a few beta testers.

Steph

Apr 02, 2008, 04:37 PM

Following this discussion I think I've realized the real reason we'll never get the source code. There's such a huge team of talent here, so many of whom are willing to work obsessively for no material award. They're scared to death that what we make will completely blow them out of the water; followed by a competitor head-hunting the major contributors, or even a VC offering to setup a new company for a generous share of the profits. And once the code is floating around, there'll be multiple versions, ever expanding & changing, just like with Linux. They're probably afraid they'll permanently lose market share to a better game made by CFC's denizens & given away for free. Why buy civ5++ when we can produce any graphics we want on top of a game engine we can tweak ourselves? So they'll act like the proverbial "dog in the manger" forever.
Actually, this is part of the discussion with Take2. We need to find a solution where they keep some control, but don't commit resource...
And that's why they won't release the source code publicly.

However, just think a bit.
Solution 1) They do nothing. People get mad and program a competitor game.
Solution 2) They agree to transfer the code to the people who would have got mad... They become busy to expand CivIII, for Take2 profits... and are not working on a competitor game.

Guess what I'd do if I were Take2?

But they may not be cynical enough.

Steph

Apr 02, 2008, 04:44 PM

But.. Steph, Darque and some of their friends are making "our own civ". Ten times better, tho.

And it seems it takes ages, since they all have real lives to take care of, I know atleast Steph has kids. Its hard to make a game for free, and when the creators are all over the world. (even tho its mostly digital stuff involved).
I have three big problems to work on the game.
1) People here are really mean, and they keep releasing wonderfull new graphics. So I feel compelled to expand my mod again and again... And it sucks most of my free time.
2) I don't have that much free time between my job with lots of travel and familly time. However, it may be enough to move forward...
But there is the hardest problem...

Lack of help!

To program such a game, progammer are not enough. We need game designer who can design some detailed parts. I could do this myself, but then I wouldn't have time to program.
And most importantly, I need graphists. Granted, there are a lot of unit makers here... But that's not what I need at the moment.

For the moment, I need an artist who can make me maps graphics, as I need them. I need someone who can understand fast the tiling mechanism, send prototype graphics, and then polish them when the prototype is working.

Later, I'll need the same for the units, to see if all my ideas can work.

You see, my motivation is greatly fulled by a visual result.

If I program something, but there's nothing to see, I'm not motivated and it work slowly.
When I have graphics to play with, then I can work fast.

I can program interesting stuff real fast when I have the material I need.

So find me one or two motivated and skilled artists, and a game designer to help on the design doc, and we'll move much faster.

Steph

Apr 02, 2008, 04:49 PM

What Stormrage said. If we really had 150,000 members who wanted to see this, & each of us contributed 1 dollar/euro, I'm sure Steph could afford to quit his job & work at home. Then we'd just have to wait the same amount of time or longer, since kids are more distracting than bosses.
My kids go to school during the day, so if I could work on the program full time at home, it would move forward fast I think.

But I would not consider quitting my job for less than 50,000 € / year, and I need a guarantee of several years safe in front of me.

And that's just for me. Without the artists, other programmers, designers, testers...

To be safe, I think a team need around 100,000 € / year / person...

So if we imagine one project manager + 3 programmers + 1 game designer + 3 artists + 2 testers, that's 10 people, 1,000,000 € / year.
And I think we could need 2-3 years.

Nobody has 2 or 3 millions to invest?

Of course, this is if we can't make some profit first...

Blue Monkey

Apr 02, 2008, 05:01 PM

We need game designer who can design some detailed parts. a game designer to help on the design doc, and we'll move much faster. What would be involved (in a little more detail)? I've done some technical writing in the past & my grad school research had to do with design of VR environments for education that included looking at game & interface semiotics. The "lab manager" phase of my career involved mostly end user support, including training curriculum design & writing instructional material & manuals. I don't know if that skill set matches what you are looking for. On the other hand I don't have to work (Social Security income) & I probably should be a member of Civ Anonymous.
And most importantly, I need graphists. For the moment, I need an artist who can make me maps graphics, as I need them. I need someone who can understand fast the tiling mechanism, send prototype graphics, and then polish them when the prototype is working.When I have graphics to play with, then I can work fast.I don't know about skilled, but I have a pretty good mind for visual/spacial thinking. My only experience with terrain design was CnP (http://forums.civfanatics.com/showthread.php?t=189306). As far as it goes, I'm pretty good with CnP judging by people's comments on my icons.

I know it will frustrate Yoda Power, but I'd be willing to put off learning Poser, & even working on my own mod, if there is a realistic chance of a better game to build the mod for.

Edit: I was trained in the US Army as an Aerial Imagery Analyst/Cartographer. so add that to my skill set for terrain graphics design. & maybe someone else (Ares?) would be willing to make polished graphics later on based I prototypes made by someone like me.

Ozymandias

Apr 02, 2008, 05:20 PM

Derf

Apr 02, 2008, 06:01 PM

Looking at the thread and what they seem to have said so far, I'm not optimistic about this.
But on the other hand, I think if we sent them something like what Oz just said, we might beat them over the head with it enough to make them see reason.
Possibly.

Blue Monkey

Apr 02, 2008, 07:56 PM

2. A group of us in the U.S. (like I said earlier/elsewhere, I literally live 2 blocks from Take Two's offices) form a not-for-profit corporation whose charter is to improve Civ3 as an educational tool.Just on the off chance that this April Fools' is turning serious, as in "we are really going to do this" serious, here's a low-rez peak at my resume & samples of the kind of support I can get. I blanked out personal information for now since this is getting posted publicly.

http://img528.imageshack.us/img528/7056/p1jc9.th.jpg (http://img528.imageshack.us/my.php?image=p1jc9.jpg)http://img528.imageshack.us/img528/6270/p2ey3.th.jpg (http://img528.imageshack.us/my.php?image=p2ey3.jpg)http://img528.imageshack.us/img528/404/p3mq0.th.jpg (http://img528.imageshack.us/my.php?image=p3mq0.jpg)http://img528.imageshack.us/img528/2460/p4uj2.th.jpg (http://img528.imageshack.us/my.php?image=p4uj2.jpg)

I could probably talk at least the retired person into sitting on an advisory board in some capacity. This was my master teacher, who, after being a teacher, principal, & district administrator, was involved at the state level in curriculum design & administration. Also, a person I haven't contacted in years, but may still be around, is another of my former professors who was involved with Apple Computer's educational design efforts in the 80s & 90s.

I'm sure we've got others like myself that would show respectable credentials on the education/academic side to be members of a design team. The tech skills available are in no doubt. & I wouldn't limit the educator part of the team to the US - why exclude people like Plotinus & Ogedei (whom I understand to both be educators) if they want in? I could even see us writing a grant proposal to UNICEF or some other NGO.

But I would not personally put effort to asking those favors unless there were genuine dedication all around. That's a lot of credibility with professionals that I also care about my personal relationship with that I would be investing.

Ozymandias

Apr 02, 2008, 09:50 PM

Hi Blue Monkey,

I really don't mean to exclude anyone, especially the people you mentioned and many others. My view on having the legal entity domiciled in the U.S. is really a means to two ends: tax & liability structuring and, more importantly, giving Take Two as many reassurances as possible: their attorneys will know U.S. contract law, and having a U.S. entity guarantees that any suits they might launch (not that we would give them cause to! - I'm simply speaking to corporate assurances and comfort) wouldn't involve multiple and/or foreign entities. Of course, if we should by the Grace of Civ wind up with an NGO or some such sponsor, that's another story entirely.

Best Regards,

Oz

Blue Monkey

Apr 02, 2008, 11:58 PM

I'm in complete agreement as to the legal side, Ozymandias. I didn't mean to sound like I was arguing. I should have said more directly that it might actually be a "selling point" that we could produce something usable world-wide. In another thread I mentioned the possibility of some how tying in to the One Laptop Per Child (http://en.wikipedia.org/wiki/One_Laptop_per_Child) project or a similar non-profit education oriented project.

Mithadan

Apr 03, 2008, 01:08 AM

I don't see what good the following factoid would do, but I am also an 'educator' -- I'll be teaching two courses in Science, Technology & Society next year (on top of the regular philosophy & theology stuff), but damned if I'm going to use CivIII in my lectures! :D Interesting thought, though. Hmmmm...

Steph

Apr 03, 2008, 03:08 AM

What would be involved (in a little more detail)? .
For instance, in the design doc I have so far, I can write something like:
"When a unit is moving, it can suffer casualties because of terrain".
And later , in the terrain section
"Each terrain can has an effect on attrition".

The game designer's role would be to:
1) Be sure that the different parts of the design documents are consistant with each other, ie in the above case be sure the term attrition is used everywhere and has the same meaning (so "when a unit is moving, it can suffer attrition).
2) Write the details for some generic idea. In the above example, explain how attrition is working exactly, what are the parameters that can affect it, etc.

It needs to be relatively "abstract", so the modders can later modify the behaviour easily to fit their own mod. So I don't need something likes "in jungle, attrition is 10%, in desert is 15%", but more "Each turn, a unit has a chance to lose hit point. Chance is (terrain base attrition - unit foraging skill) * distance to base".

That is the first part of the job

A second part could be to use the different editors to create a "base game", i.e. setting jungle attrition to 10%, etc. to check that the model is working, and propose a default game.

The modders around the world would then take care of all the scenarios :D

I don't know about skilled, but I have a pretty good mind for visual/spacial thinking. My only experience with terrain design was CnP. As far as it goes, I'm pretty good with CnP judging by people's comments on my icons.

I know it will frustrate Yoda Power, but I'd be willing to put off learning Poser, & even working on my own mod, if there is a realistic chance of a better game to build the mod for.

Edit: I was trained in the US Army as an Aerial Imagery Analyst/Cartographer. so add that to my skill set for terrain graphics design. & maybe someone else (Ares?) would be willing to make polished graphics later on based I prototypes made by someone like me.

I have designed a new tiling system. Quite different from Civ3. Allowing an easy way to add new terrain, with possibly dozens of variants, and animatd maps.

Some copy & paste work can be enough to start with, but I think using a 3D tool to model the terrain and apply proper texture would be better.

Basically, I need someone who can make little tiles as in the top of the picture below(they are independant images, in PNG, they can be 256 colours indexed, or true colour with full alpha transparency). The most important part is they need to be seamless: be sure that the edge of a cliff tile are connected properly to another cliff tile.

Then the tiles are assembled to give the kind of small map you see below. The tiles in this screenshot I made myself from Bhiita's initial work, but I don't have time nor image editing skill to make them perfect, they are just a "mock up". So it's quite uglly: there is no texture, no wave in the sea, no effect... It doesn't tile very well, and of course it's not animated.

http://img387.imageshack.us/img387/6739/coastsgfad6.jpg

I need this for several terrains, coasts... And later similar tiling for relief, rivers, vegetation, etc.

Steph

Apr 03, 2008, 03:15 AM

Steph - you make it sound like the following might be a viable scenario:
1. A group of us review the code/APIs/whatever under non-disclosure to see how much of it (AI handling artillery and ships; etc.) has to be gutted, and if what's left over is worth working on (as opposed to simply starting from scratch).

Of course, the first step would be for us to be able to assess the status of the code. Is it well organized? Is it well commented? What is the available documentation? And from there, to see it is possible to correct/expand easily, or if it's to messy / complicated and starting from scratch would be better.

I would refuse to go further until we can do this.

2. A group of us in the U.S. (like I said earlier/elsewhere, I literally live 2 blocks from Take Two's offices) form a not-for-profit corporation whose charter is to improve Civ3 as an educational tool.

3. This entity is granted free license by Take Two to further develop C3 strictly for non-commercial (remember that educational part?) purposes.

Let's playing devil's advocate...
I am a Take2 executive. The goal of my company is to make profit. How do you convince me you "non-commerical" purpose will help me make profit?
...
I don't think it's the right approach.

The right approach would be to find a solution were Take2 benefits and make profit.

However, if we can find some common ground, we still need a legal frame to negociate with them, and I agree it would be better to have an official structure based on American law. It's with the "non for profit" part I don't fully agree, because it may not be well received by Take2.

We need a structure where we can work for non personnal profit (although I wouldn't mind a little extra, for instance to be able to upgrade my computer and work faster), but that sill give the possibility for Take2 to make additional profit.

Weasel Op

Apr 03, 2008, 08:43 AM

As a full time college student, I support the use of Civ3 for educational purposes :D

Seriously though... I'm glad this discussion got kicked off again. I'm fairly cynical about the whole thing (I've mentioned before that a community-built Civ3 clone would probably be more feasible, though that presents its own problems) but there's no reason not to try.

I'll be graduating with a BS in Software Engineering in just over a year, if that's at all applicable. I would be interested in helping out with design documentation, either for Civ3 code or SSS, but I probably won't have the time to commit to it in the near future. Remember what happened last time I volunteered for SSS!- though admittedly I jumped in over my head on that one. Documentation is something that I actually have some experience in, and am quite good at if my analysis & design course is any indication.

Blue Monkey

Apr 03, 2008, 12:47 PM

I'm confident I could do the kind of design document writing & editing you posted about: take your raw notes, turn them into consistent vocabulary, then translate the terms into guidelines/parameters programmers can use to write the specific code from. A lot of what I did when I worked at the Polytechnic University was function as a "translator" between the computer scientists & programmers and the professors/clients from the humanities: Take vague outcome desires & turn them into requests for something programmable, and turn the programmers questions back into something the clients could answer without understanding anything about look-up tables, iterative loops, yada yada... That kind of work sounds like something I could do while still making things here. And alpha testing to make sure the game functions is no problem either. I've also done a little code documentation. I can write specifically enough for programmers to track down specific bugs.

Edit: The terrain on the other hand - sounds like you need someone with 3D experience. that's not me.

Tank_Guy#3

Apr 03, 2008, 01:12 PM

If I had any programming skill at all I would volunteer for this. However, I struggled with entry level C++ and Visual Basic (2005 if I remember correctly), so it would probably be best if I didn't.

I could ask my friend who is more programming savvy if he'd be interested in helping, but right now I know for a fact that he's loaded to the gills with work and family stuff. A Wisconsin version of Steph :lol:

@ Blue Monkey: shot you a PM also.

Blue Monkey

Apr 03, 2008, 04:22 PM

@ Blue Monkey: shot you a PM also.Got a similar PM from Stormrage. Thank you both for that. Yes my name is not blocked out in one of the letters. That's okay. The name's common enough - I've met unrelated people with mine, my brother's, and an uncle's full name. I used to enjoy getting a lieutenant's paycheck as a private, except that he was already waiting at the post box every single month with my meager pay! :lol: My primary concern was revealing others' identity without permission + addresses & other personal info. Thanks for the concern.

Keroro

Apr 03, 2008, 06:07 PM

I'm keen to see this all move forward if possible. I still look to the Call to Power II code release as an ideal - I am certain that the source code release has helped CtoP II sales. The important thing with that release is that no release could be used without the original CtoP II CD in the computer's drive. This is essential as far as I can see as it guarantees Firaxis/Take 2 that they won't lose money from this enterprise, even if they may not make a huge amount from it either.

Steph - I might be able to help out with the 'game designer' role you mention - shoot me something by PM if you want me to take a look at some files. Of course it might take a while to get used to what you're trying to achieve / what you need to be able to progress, but I'll do my best.

Ozymandias

Apr 03, 2008, 07:02 PM

Okay, everybody take a deep breath. What I'm about to write is based on over 25 years in software development.

Programmers program from what are generally called "technical specifications" which comprise all the Rules their code is to encompass.

"Technical specifications" are derived from "functional specification": what is the step-by-step logic to be followed.

I write this having heard of the CTP2 AI rewrite; I don't say our effort cannot be done. But the aspect which troubles me most is the "functional specifications".

I cannot emphasize enough that these are Rules, if-condition-x-do-action-A-else-do-action-B.

We've all complained about how the AI is fairly idiotic in (I choose at random) its use of artillery.

I challenge anyone to write the functional specifications for what the AI is to do, step-by-step, with a newly built artillery unit (and I've already made it easy by ommitting how the AI would decide to build a unit):

For simplicity's sake, imagine a 10x10 grid with a friendly and an enemy city on either side (the artillery has obviously just been built in the friendly city).
There is 1 defender in the friendly city, and three units are one square away from the enemy city, which has two defenders.
In between are two enemy and two friendly light cavalry units.
All terrain is grassland.

Kindly write out the decisions the AI must make for the artillery unit in question for just three moves; kindly recall that computers don't understand vagaries, so all factors must be taken into account and the AI's response clearly stated.

I'm not saying it can't be done - far from it (indeed, I can imagine Vingrjoe chomping at the bit to rewrite the naval rules ;) ). I just want to call everyone's intention to the kind of work which will be the core efforts to improve the game beyond the "look and feel".

If we can muster six people who can write just such descriptive logic, then I'd say we have a very good shot at turning out a fine product.

My Best Regards To All,

Oz

Stormrage

Apr 03, 2008, 07:31 PM

Wait, just write the rules in plain english, or atleast engrish, or write them in a programming language?

Ozymandias

Apr 03, 2008, 07:36 PM

Wait, just write the rules in plain english, or atleast engrish, or write them in a programming language?

Plain ol' English. The point here is the detailing of the logic - the set of rules - the AI would employ to perform this "simple" task.

-Oz

Ozymandias

Apr 03, 2008, 07:42 PM

I don't think it's the right approach.

The right approach would be to find a solution were Take2 benefits and make profit.

However, if we can find some common ground, we still need a legal frame to negociate with them, and I agree it would be better to have an official structure based on American law. It's with the "non for profit" part I don't fully agree, because it may not be well received by Take2.

We need a structure where we can work for non personnal profit (although I wouldn't mind a little extra, for instance to be able to upgrade my computer and work faster), but that sill give the possibility for Take2 to make additional profit.

"Not for profit" in the US essentially means that any extra monies existing after expenses be rolled back into the enterprise rather than disbursed as dividends, etc. Salaries, hypothetical payments to Take Two, etc. can still be paid. It is also a tax advantageous structure.

Best,

Oz

Stormrage

Apr 03, 2008, 07:43 PM

3 moves? But there are so many possibilities! I guess there is your point..

Blue Monkey

Apr 03, 2008, 07:48 PM

Wait, just write the rules in plain english, or atleast engrish, or write them in a programming language?Write the rules explicitly enough in English that programmers can sit down to write the actual code without having to figure out the decisions the AI must make.

One little tiny piece of the "rules" might be "barrage the cavalry unless the infantry has a higher attack value and can attack the city on next turn". But in the little, simple situation he describes there might be dozens of deciding factors that have to be weighted in the rules (such as the move, attack, defense values of every single unit and how those are changed by terrain, fortifications, type of opponent), then translated to code. Then there have to be rules to make decisions about how that situation affects others on the map at the same time, what order to do individual actions in, etc.

And to re-emphasize what he is saying a design team has to write out all the rules; the programming team just turns them into code. Which it would be far faster to work from the existing game engine, and also why Steph hasn't popped out his game as quickly as Sandris pops out units.

Ozymandias

Apr 03, 2008, 07:48 PM

Plain ol' English. The point here is the detailing of the logic - the set of rules - the AI would employ to perform this "simple" task.

-Oz

Example:

1. Turn one: Is there an enemy unit within striking range? If yes, do not move; if no, move one tile towards city.

2. Turn two: Is there an enemy unit within striking range? If yes, retreat to city UNLESS another unit has been built in city THEN do not move artillery AND have newly built unit move to guard artillery.

Got the idea?

Best,

Oz

Ozymandias

Apr 03, 2008, 07:50 PM

Cross-post with Blue Monkey; his explanation is spot-on.

-Oz

Stormrage

Apr 03, 2008, 07:52 PM

But thats not Sparta, thats Madness! Kudos to all the people out there doing these things.. Major kudos.

Blue Monkey

Apr 03, 2008, 07:56 PM

Thanks for the stamp of approval, Ozymandias; I'm only a jargon translator. It's amazing that the people who do that program/design stuff for a living ever have the time to post here, the way that Brad Oliver used to help us on the Mac forum.

Ozymandias

Apr 03, 2008, 07:59 PM

@Blue Monkey - "only" a jargon translator is to do yourself an injustice. I used to joke on Wall Street that I spoke three languages: English, Finance, and Technology; it served me well.

Best,

Oz

Blue Monkey

Apr 03, 2008, 08:05 PM

Yeah, there's a reason my occupation is listed as "itinerant semiotician" so often.

Ozymandias

Apr 03, 2008, 08:07 PM

Yeah, there's a reason my occupation is listed as "itinerant semiotician" so often.

Yes, I've been trying for months now to work the distinction between la langue and la parole into a witticism. :)

Blue Monkey

Apr 03, 2008, 08:12 PM

Yes, I've been trying for months now to work the distinction between la langue and la parole into a witticism. :) Then if you told it to yourself, would you understand it well enough to write it down for the rest of us? :lol: The real trick would be to work it into a Japanese FPS based on a Peter Greenaway movie.

Ozymandias

Apr 03, 2008, 08:18 PM

Then if you told it to yourself, would you understand it well enough to write it down for the rest of us? :lol: The real trick would be to work it into a Japanese FPS based on a Peter Greenaway movie.

Well, maybe if I voluntarily had my corpus collosum split ... nah ... but the soundtrack to The Draughtsman's Contract might work alarmingly well for a manga-based FPS. ;)

Weasel Op

Apr 03, 2008, 11:05 PM

Of course, we don't even know how the AI is programmed- whether coded in with the rest of the game, or with scripts, or rule-based, or some other implementation. :ack:

Steph

Apr 04, 2008, 02:10 AM

1. Turn one: Is there an enemy unit within striking range? If yes, do not move; if no, move one tile towards city.

2. Turn two: Is there an enemy unit within striking range? If yes, retreat to city UNLESS another unit has been built in city THEN do not move artillery AND have newly built unit move to guard artillery.

If you really want a smart A.I. it's much more complicated.
Because you need to weight dozens of parameters:
- The presence of an ennemy unit, and friendly unit, and bonuses.
Ex for turn two:
If ennemy unit within striking range, and the city defense bonus is important (walls, etc), it may be wiser to move to the city for a better defense than moving the unit from the city.
Or if a cavalry unit is not far away, it may be better to bombard with the artillery to weaken the ennemy unit, and then attack with the cavalry, etc.

And also, you have to take into account a "grand strategy" versus "local tactics"...

Steph

Apr 04, 2008, 02:31 AM

But in the little, simple situation he describes there might be dozens of deciding factors that have to be weighted in the rules (such as the move, attack, defense values of every single unit and how those are changed by terrain, fortifications, type of opponent), then translated to code. Then there have to be rules to make decisions about how that situation affects others on the map at the same time, what order to do individual actions in, etc.

And to re-emphasize what he is saying a design team has to write out all the rules; the programming team just turns them into code. Which it would be far faster to work from the existing game engine, and also why Steph hasn't popped out his game as quickly as Sandris pops out units.
That is step 5!

Here is how I think it should go for a new game. I'll use just the example of the battle.

Step 1: Design the general principle. Ex: A unit can have a weapon and an armor. The weapon has an attack strength, the armor a defense strength, etc. and list the parameters that you think should be taken into account (terrain, morale, range, speed, etc).

Step 2: Abstract these parameters and try to think one step toward "moddability". We want a very moddable engine, so we should not have something to strict. We need to be able to allow space scenario, fantasy scenario with magic, etc.

Step 3: Model the game engine that allows this moddability and contains the designed parameters.

Step 4: Check that the engine is working fine with human to human game.

Step 5: Design the AI.

To illustrate the point:
- Unit have an attack and a defense. Chance to hit = A / ( A + D).
- Test the rule is working and give good result in game. Then you may realize you need to introduce a new rule : unit type so pikemen can have a bonus against cavalry.
- When satisfied with the mechanism, where you can use pikeman first against cavalry.

We need a design team and programmer team that are working in close relationship, so the design team propose a solution, the programmers prototype it, we test, update, and repeat.

Ozymandias

Apr 04, 2008, 09:27 AM

@ Steph,

The complicated nature of the beast was precisely my point.

As for development methodology, I'm fairly agnostic about that, as long as we DON'T use Microsoft's! (They don't believe in a Project Manager role, i.e., someone at the hub of all issues making certain everything and everyone is working together).

Best,

Oz

Ozymandias

Apr 15, 2008, 06:44 PM

:hammer:

As no one has taken up my challenge, I would like to suggest that it is conceivable to change quite a few gameplay details without going anywhere near the AI.

Some random thoughts:

1. Being able to set a limit on RR travel. Limiting certain units to RRs (e.g., armored trains).
2. Moving rivers back to the center of tiles (a la C2) which IMHO is much more realistic - and specifying that units (e.g., gunboats) could be limited to rivers.
3. A pet desire of mine: to set how far away from a city a unit can move, and have this limitation be changeable and/or eliminated by a tech.

I'm sure you all get the idea.

- And, yes, for ye olde timers :old: Zulu did compile such a list long ago and far away, combining both desired non-AI and AI changes.

Let me know what you think.

Best,

Oz

Mithadan

Apr 15, 2008, 08:11 PM

I'm a big fan of each of those three.

Steph

Apr 16, 2008, 11:51 AM

:hammer:

1. Being able to set a limit on RR travel. Limiting certain units to RRs (e.g., armored trains).
2. Moving rivers back to the center of tiles (a la C2) which IMHO is much more realistic - and specifying that units (e.g., gunboats) could be limited to rivers.
3. A pet desire of mine: to set how far away from a city a unit can move, and have this limitation be changeable and/or eliminated by a tech.

1 is nice.
3 is a must have.
2 I'm not sure, as having rivers on the edge can make interesting defensive lines.
So are river more important as a defensive line and borders, or as waterways? Personnaly, I prefer them as defensive lines.

Quinzy

Apr 16, 2008, 12:02 PM

I'm with Steph on the rivers. Perhaps have both options?

Civinator

Apr 16, 2008, 12:13 PM

So are river more important as a defensive line and borders, or as waterways?

In my eyes both is important. Set the rivers in the center of the tile, make it shipable and allow to cross the tile with the river only for certain units, p.e. settlers, combat engineers and so on who can build bridges.

The comeback of some simple Civ 2-flags would be nice too.

The "ignore-walls"-flag in civ 2 is a good flag for assault guns (Sturmgeschütze), that could be very useful for Civ 3 too.

The Civ 2 "pikeman-flag" is a classic. I think no more words here are needed.

The "ignore-impassable-terrain"-flag of Civ 2 ToT could also be very interesting (p.e. for amphibious operations in scenarios).

Ozymandias

Apr 16, 2008, 03:51 PM

Rivers as defensive boundaries can be easily addressed by giving units occupying river tiles a negative defense bonus. I'm also definitely in agreement with Civinator on the good ol' C2 flags; this was partly why (when this first all came up) the first thing I wanted to look up were the programming APIs (if any!) and not necessarilly the mainline code itself.

Best,

Oz

Steph

Apr 16, 2008, 04:07 PM

Rivers as defensive boundaries can be easily addressed by giving units occupying river tiles a negative defense bonus.

A negative defense bonus? Did you mean a positive?
beside, it's not having a river in the tile that should give a bonus. it's having the river between you and the attacking army.

Ozymandias

Apr 16, 2008, 07:32 PM

A negative defense bonus? Did you mean a positive?
beside, it's not having a river in the tile that should give a bonus. it's having the river between you and the attacking army.

Assuming the river is running through the tile, any force in the tile should be considered split on both banks; in the act of crossing; etc. A negative bonus allows a defensive line to be formed in tiles adjacent to said river with the force in the river tile at an obvious disadvantage. This IMO is an adequate representation of a defensive line behind a river.

Best,

Oz

Steph

Apr 17, 2008, 02:00 AM

Assuming the river is running through the tile, any force in the tile should be considered split on both banks; in the act of crossing; etc. A negative bonus allows a defensive line to be formed in tiles adjacent to said river with the force in the river tile at an obvious disadvantage. This IMO is an adequate representation of a defensive line behind a river.

This would be true only if the river occupies most of the tile, i.e. a unit in the tile is actually wading it or swimming.
Let's consider a world map. With 400x400 tiles.
Earth is 40,000 km circumference. So a tile is 100kmx100km.
I don't know many rivers which are 100 km wide.

Beside, the bonus should apply when actually crossing a river. Imagine the river is running North South. The bonus should apply when an army is on the East bank, and try to attack an army on the West bank.
It would not apply to a unit attacking from the North or South on the same bank as the defending unit.

Third, I don't see why a commander would be stupid enough to split an army in two and put it both side of a river when in a defensive position. This happens only temporarily when trying to cross the river.

So, with your method:
Unit A (attacker) wants to attack unit D (defender).
D must be in a tile adjacent to the river tile. So 50km from the river with my world map example.
Unit A goes to the river tile. The little guys are fording / wading / swimming, etc.
Next turn : Unit A will attack unit D. As civ doesn't take into account the terrain in the tile of the attacker, unit D has no defensive bonus. Which negate a bit the interest of the rivers as defensive position.
Or unit D can attack unit A while it is still in the river. But then unit D is an attacker, not a defender... And the A/D value of the units are not the same.
And what happens if unit A as enough movement points to go in the river tile and then attack unit D in the same turn?

I'm sorry to say so, but your idea is a very inadequate representation of a defensive line behind a river.

Bjornlo

Apr 17, 2008, 07:03 AM

I'm sorry to say so, but your idea is a very inadequate representation of a defensive line behind a river.

I disagree. If the river square is a negative modifier this models rather nicely both the effect of defending across a river since no the modifier will be applied to any and all facings so long as the attacker is "crossing" a river tile. This would be true no matter where the river is. How is this less accurate than attacking across a river running down the seams of the tiles in every concievable scenario.

I see no downside. The upside is it would allow for river only units. These units could have zone of control to snipe at units and could have their relative strength adjusted if needed. Even cooler would be the coastal and river only units. But this would neccessitate bumping all naval units A/D a bit (since the river unit might be bumped up to cancel out the minus ground units have when on a river tile. It also opens up river barges to move units up and down rivers, etc....

I really like the idea of going back to C2 style rivers.

Steph

Apr 17, 2008, 07:21 AM

I disagree. If the river square is a negative modifier this models rather nicely both the effect of defending across a river since no the modifier will be applied to any and all facings so long as the attacker is "crossing" a river tile. This would be true no matter where the river is. How is this less accurate than attacking across a river running down the seams of the tiles in every concievable scenario.

I wonder if you read what I wrote.
Attacker A is your "river" tile. Defender D is adjacent to the river tile.
A attacks D : as the combat model doesn't take into account the terrain in the attacker but only the terrain in the defender tile, the river has no effect for defense.

And if defender D attacks attacker A while it is still in the river tile, then it's an attack, not a defense, and the river gives a disadvantage to the defender.

This system is screwed on so many levels I'm amazed that you don't see it.

Ozymandias

Apr 17, 2008, 11:09 AM

Hi Steph,

I don't disagree with your analysis (although, insofar as each tile is 100 km wide, one wonders about weapons with range >1 in general).

My main point is that the advantages of having rivers running through tiles (gunboats; early game rapid movement; cities acting as bridges; etc.) far outweighs the negatives of rivers running betwen tiles, especially as I've yet to see the AI take advantage of a river defense line of any length! (Hm, I suppose I'm essentially echoing Bjornlo; c'est la guerre.)

However, I'm not sure if I'm following your point (it's just a bit convoluted, keeping track of the respective units) :

"And if defender D attacks attacker A while it is still in the river tile, then it's an attack, not a defense, and the river gives a disadvantage to the defender."

So if Unit1 attacks Unit2 while Unit1 is in the river tile, neither side gains an advantage or disadvantage from the river.
If Unit2 attacks Unit1 while Unit2 is in the river tile, neither side gains an advantage or disadvantage from the river.
The only time terrain comes into play is if Unit1 attacks Unit2 while Unit2 is in the river tile, then Unit2 is at a disadvantage.

Again, by no means perfect, but, then again, what in Civ is?

Best,

Oz

Steph

Apr 17, 2008, 01:19 PM

I don't disagree with your analysis (although, insofar as each tile is 100 km wide, one wonders about weapons with range >1 in general).

My main point is that the advantages of having rivers running through tiles (gunboats; early game rapid movement; cities acting as bridges; etc.)

Why should city act as a bridge? Rivers were often a protection, with city on an island, in a curve, or only one side, or two cities on each side, which later grew an joined together to form a larger city.
There is no reason a city creates automatically a bridge, except if there is somewhere to go on the other side of the river. Which is also present in game: build a road on the other side so you have somewhere to go.

far outweighs the negatives of rivers running betwen tiles, especially as I've yet to see the AI take advantage of a river defense line of any length! (Hm, I suppose I'm essentially echoing Bjornlo; c'est la guerre.)
Your opinion, but I don't share it. The problem is you are thinking of a solution for a small scale scenario, not for a large scale strategy game.
Your "river" square is for a tactical game, not a strategical one.
And if we have access to the source code, wouldn't it be better to fix the AI so it can use a defensive river line correctly, rather than to change the way rivers are working, and make a new AI to use it correctly?

"And if defender D attacks attacker A while it is still in the river tile, then it's an attack, not a defense, and the river gives a disadvantage to the defender."

So if Unit1 attacks Unit2 while Unit1 is in the river tile, neither side gains an advantage or disadvantage from the river.

Exactly... So how do you create a defensive line with a river, as you don't get a benefit from the river?

If Unit2 attacks Unit1 while Unit2 is in the river tile, neither side gains an advantage or disadvantage from the river.

Same case as above.

The only time terrain comes into play is if Unit1 attacks Unit2 while Unit2 is in the river tile, then Unit2 is at a disadvantage.

Ok, so how does river allow defensive line, of they only give disadvtange to a defending unit, and no advantage to an attacking it?

Beside... Do you imagine a unit just sitting there, in the middle of a river... During one full turn, 20 years, just waiting to be attacked?

A river is used as a defensive line to entrench on the other side, control, the fords, the bridges... And it just thus give a defensive bonus to a unit when attacked by a unit croming form across the river.

Again, by no means perfect

Ah, at last we can agree on something :p

Civinator

Apr 17, 2008, 02:15 PM

Interesting discussions, but mostly about things that can be done with the existing editor.

The new proposal was to give the rivers back in the center of a tile as it was done with Civ 2. I completly agree with that proposal.

About the discussion about rivers as defense lines, I also start with a handling of the editor that still can be done. In SOE for some time I seriously reflected to set the jungle-terrain graphics to transparency and use this green terrain as a special terrain along rivers. The terrain is impassable for wheeled units and all units have the wheeled flag, except combat engineers (workers) and some amphibious tanks. The combat engineers can move on that river-terrain and can build bridges for the other units. At some special tiles there is terrain, that can be passed by units with the "wheeled flag". On that terrain bridges over the river (resource graphics) can be posted. So the river even with the existing editor can become a real obstacle. At last the rivers in SOE were not set in this way, as the terrain was needed for something more spectacular.

Now, if we could work with the source code and have the "cross impassable"-flag, the river-terrain can be set to impassable and the worker-units and amphibious tanks could cross that terrain and the "wheeled-flag" could be used for other interesting duties

Ozymandias

Apr 17, 2008, 04:22 PM

Why should city act as a bridge? Rivers were often a protection, with city on an island, in a curve, or only one side, or two cities on each side, which later grew an joined together to form a larger city.
There is no reason a city creates automatically a bridge, except if there is somewhere to go on the other side of the river. Which is also present in game: build a road on the other side so you have somewhere to go.

Yes, but any city built astride or upon a river can be depended upon to have a boat ferry which can, at Civ scale, be considered to approximate a bridge.

Your opinion, but I don't share it. The problem is you are thinking of a solution for a small scale scenario, not for a large scale strategy game. Your "river" square is for a tactical game, not a strategical one. And if we have access to the source code, wouldn't it be better to fix the AI so it can use a defensive river line correctly, rather than to change the way rivers are working, and make a new AI to use it correctly?

Actually I am thinking strategically: movement along rivers etc. vs. one particular defensive formation which the AI currently does not use, which would be used in a battle ("tactical") - and not having to write an AI routine to utilize river line defenses (again, my opening salvo for this particular round was to recognize how difficult AI programming is and what gameplay improvements can be made without AI modification)

Besides... Do you imagine a unit just sitting there, in the middle of a river... During one full turn, 20 years, just waiting to be attacked?

No, but neither do I imagine defensive units fortified in place for a couple of centuries. :D

A river is used as a defensive line to entrench on the other side, control, the fords, the bridges... And it just thus give a defensive bonus to a unit when attacked by a unit croming form across the river.

Ah, at last we can agree on something :p

:lol: Most likely quite a bit. :)

Best,

Oz

Steph

Apr 18, 2008, 01:43 AM

Yes, but any city built astride or upon a river can be depended upon to have a boat ferry which can, at Civ scale, be considered to approximate a bridge.

Yes, it could, but I think building a road on the other side of the river, is enough to simulate that effect. Usually, you create roads to connect cities anyway.

Actually I am thinking strategically: movement along rivers etc.

I agree that this parts could be reworked, but this could be done by keeping the river on the edge. If moving from tile A to tile B, and a river exists on the edge of tile A and B (of course on the same edge), then consider it as a "waterway" that could increase movement speed. But only when following the river, not when crossing it.

vs. one particular defensive formation which the AI currently does not use, which would be used in a battle ("tactical") - and not having to write an AI routine to utilize river line defenses (again, my opening salvo for this particular round was to recognize how difficult AI programming is and what gameplay improvements can be made without AI modification)

First, I don't believe in having the AI able to use river properly for defense if it is not upgraded. Whatever the method used (between tiles or in tiles), the AI will not use it correctly. Unless we manage to correct the AI.
Secondly, I think you are mixing two things. You are not using the strategical and tactical abstractions of the game mechanism correctly.

In a war, when a river is in use, the defending army will fortify on one side of the river, hold the bridge, the fords, etc. The attacking army needs to find a weak spot, or concentrate enough force to break through, establish a beach head and then expand from that. Such an attack can be costly, as the defender as a good terrain advantage.
I think this is best represented by giving a bonus to the defender: if it is attacked, the attacker is likely to suffer more losses.
I really don't see how giving a negative bonus to a unit in the river can simulate this, even remotely.
This is possible only in tactical level wargames, with a river that is large, with reactive fire from the shore, etc. This things are not available in civ "epic game". I agree it could be useful for some scenarios with a much smaller scale.

Now, about strategical and tactical: the strategical part is to put your forces on one side of the river. Large river like the Rhine for instance. And then sit there, holding the bridges, until attacked. Then the tactical part is for the attacker to try to cross somewhere (concentrating on one point, or probing several points at the same time, but it doesn't matter). We don't need to know how he crosses (bridge? Ford? Small boats?), nor how exactly it gives a bonus to the defender (shoot at the attacker while they wade in the river, sink the boats they are using, form a wall of pikes at the end of the bridge that is difficult to turn?). We just need "crossing river = more loses for the attacker".

In other word, the tactical details of the difficulty in the river crossing is abstracted on the strategical level as "bonus for defender if attacker cross the river".
The strategical abstraction simulates the expected tactical effect.

But with your solution of a negative defending bonus for a unit in the river tile?
I'm still waiting for a clear illustrated explanation of how it is supposed to simulate a tactical or strategical reallity (and I'm speaking only of the battle mechanism part here, not of the economical or fast exploration part).

No, but neither do I imagine defensive units fortified in place for a couple of centuries. :D

I don't have real difficulty with that on a strategical abstraction level.
The unit is send to the river, establish small strongpoints to hold the bridges, fords, communication network, forces in reserve to quickly reinforce a strongpoint when it's attacked, training to learn the terrain, etc.
Of course, it's not always the same men that stay here for 20 years. But new recruits come, are trained to know the procedures set in place...
All this abstracted by "fortify behind the river".