1.03 status update for June 17th!

[MarshallEllis]

Hey guys:

Just a quick FYI that I am trying to improve the AI and only stopping this work in case of a crash bug.

I still do not have a schedule for posting the first BETA but will keep you updated!

Now, back to work...

[Pans]

this is really good news.
I´m just have a game running with 1.02k where Austria (human) is at war with Turkey (KI). The Austrian forces directly went to Constantinople and staying there for 4 Month. Turkey were at full strengh and was not moving at all... not a single move.
By the way... same for Russia - DOW to Sweden but no move into Finland anyhow. Corps still standing with Bagration on the border?!

Hope this will be better with 1.03

cheers
Andreas

[Jimmer]

I would highly recommend small patch changes done more often. Since release, there have been only two patch levels that actually got released. That's over 6 months. That's WAY too long for patches.
 
I think you should shoot for releasing at LEAST every month, until the game is stable and solid.
 
To do that, you need to cut down drastically on what is included in each patch. Bite the bullet on a certain date, and refuse to add ANYTHING more, with the sole exception being code to fix any bugs that the patch itself caused or brought to the forefront. Make no other exceptions once the date is crossed, even for game-stopping critical bugs people find in the released version.
 
Then, go one step further: A second date, after the first, means that nothing more goes in except code to fix game stopping new bugs that this patch caused.
 
If you ran on a monthly schedule, the dates would probably be at the end of second and third weeks.
 
First two weeks: Include all code that is deemed completely ready for production. If it's not done, it stays out, no matter what. Only game-stopping bugs breach this rule.
 
Third week, the first date is passed. No new code goes in except code to address problems the patch itself created or brought into the foreground.
 
Fourth week, limit changes to show-stopping bugs only. And, even for show-stopping bugs, address them ONLY if they were created or uncovered by the patch. (The idea of "uncovered" is to prevent regressions: Having this release be worse than the last, in any area.)
 
Such a schedule would mean two things:
 
First, you (the programmer) will have at least two trains of code changes to keep track of: The one for the current patch level, and the one for future patches.
 
Second, you will have to have plugged ears to put up with the screams. People always seem to think THEIR bug is the most important one (by "their", I mean affecting their current game). These will have to be ignored. But, I believe most people will accept it if they know that their fix should be in next MONTH'S release (rather than the next one, whenever).
 
I've got another suggestion, but I'll post that separately, so they can be logically discreet.

[Jimmer]

My other suggestion is to do AI changes in small steps. Yes, people will scream, but not when some of their pet tricks to kill the AI stops working. I know you had a suggestion to change GB's naval to make sure it blockades every French port. There are still loopholes, but this made the French player have to think some in order to bury GB early.
 
If the next AI enhancement caused GB to put the right number of ships in each port, some of THOSE loopholes will be filled, etc.
 
The AI does not need monumental corrections. It needs lots of little changes. Lots of changes translates into lots of bugs, which means longer release cycles. IMO, this game cannot afford delays, especially long ones. The natives are getting restless.
 
So, include SOME AI changes each time you make a release. Limit yourself to only that which is complete when the month starts, no matter how much you REALLY want to get that "one little change" into the code. If you follow the release cycle model I posted just above this, you'll find that the AI will be making small, but relatively constant, changes to its thinking. That would keep gamers on their toes, at a minimum.
 
Examples, told in anecdotal form:
 
"WHAT? Russia actually ATTACKED my Turkish corps scattered around next door, during winter?"
 
That ONE change would cause most players to quit and restart (as Turkey). And, in the new game, they would certainly not make that mistake again.
 
 
"Hey, how come France now has 10 different starting setups for naval and 10 for land, and he can mix and match them to make 100 different starting positions?"
 
This change would force a human GB to be MUCH more intelligent about her blocades.
 
 
"When did Prussia get enough brains to actually attack some Germanic minors?"
 
Oh, sorry, that one is already in place.
 
There are HUNDREDS of such changes you could make. All small, but each one will cause the human player to do a backpedal. Yes, it will still be easy to beat the AI, but they'll have to actually think about it. And, as time goes on, it will continue to improve. Slowly, but surely.
 
Even now, with some of the very minor changes you've made, I've found the AI to be slightly less of a pushover in certain areas. That's the idea I'm talking about: Small changes over time. Shake up the players. Give new players a little more of a challenge. But, keep growing the AI's intelligence.

[delatbabel]

ORIGINAL: Jimmer
"When did Prussia get enough brains to actually attack some Germanic minors?"

Shows you what differences there are between what people think of as "intelligence"!

As Prussia, I never attack German minors. If you attack and conquer a minor you get -1PP for attacking it, +1PP for conquering it. Net gain, zero. When France goes on the rampage and takes them all away from you, you lose -1PP for each one. Net loss, quite a bit. Plus the German minors work as a nice "buffer" zone that France has to conquer and then defend to crawl over to hurt you, and by the time that's done you probably have some nice Russian corps and Bagration sitting in Berlin waiting for him, as well as Charles rampaging through Northern Italy if you've planned things right.

So leave the minors neutral I say. Let France spend the PPs on them.

[bresh]

ORIGINAL: delatbabel

ORIGINAL: Jimmer
"When did Prussia get enough brains to actually attack some Germanic minors?"

Shows you what differences there are between what people think of as "intelligence"!

As Prussia, I never attack German minors. If you attack and conquer a minor you get -1PP for attacking it, +1PP for conquering it. Net gain, zero. When France goes on the rampage and takes them all away from you, you lose -1PP for each one. Net loss, quite a bit. Plus the German minors work as a nice "buffer" zone that France has to conquer and then defend to crawl over to hurt you, and by the time that's done you probably have some nice Russian corps and Bagration sitting in Berlin waiting for him, as well as Charles rampaging through Northern Italy if you've planned things right.

So leave the minors neutral I say. Let France spend the PPs on them.

Well as Prussia Mecklenburg seems important. But guess conquest is not required.
But some screenings to prevent a quick French move for Berlin should be considered, and depot-garrisons are cheaper pp wise.

Regards
Bresh

[MarshallEllis]

Jimmer:
 
I agree that more releases would be better. I would love to do a release a month. That is a goal of mine. I may try to tweak the AI in 1.03 a bit then release after some good testing, possibly making the AI a goal over ther next several patches. Let me try this...
 
 

[NeverMan]

I just want to say that if you are the only developer on this game then I feel sorry for you Marshall and shame on Matrix. The game is so buggy and messed up right now that there should be a team working on this to fix this game ASAP, at LEAST to get it to a credible level. Right now, it's not. I imagine that it will probably take 1.09xx before it gets there.

Smaller patches released more frequently (which it seems is what you did with 1.02x) would be more desirable. It would be better for Matrix's reputation and for the players who paid for the game and easier on the testers and probably easier on you.

[Jimmer]

ORIGINAL: delatbabel

ORIGINAL: Jimmer
"When did Prussia get enough brains to actually attack some Germanic minors?"

Shows you what differences there are between what people think of as "intelligence"!

As Prussia, I never attack German minors. If you attack and conquer a minor you get -1PP for attacking it, +1PP for conquering it. Net gain, zero. When France goes on the rampage and takes them all away from you, you lose -1PP for each one. Net loss, quite a bit. Plus the German minors work as a nice "buffer" zone that France has to conquer and then defend to crawl over to hurt you, and by the time that's done you probably have some nice Russian corps and Bagration sitting in Berlin waiting for him, as well as Charles rampaging through Northern Italy if you've planned things right.

So leave the minors neutral I say. Let France spend the PPs on them.

As I recall, we do still get money and manpower from them, don't we? I mean, while we own them?

Also, please remember that the AI Prussia could be playing against an AI France. Don't forget to factor that in: Prussia COULD keep them the whole game (unless a human is playing France).

[Jimmer]

ORIGINAL: Marshall Ellis

Jimmer:

I agree that more releases would be better. I would love to do a release a month. That is a goal of mine. I may try to tweak the AI in 1.03 a bit then release after some good testing, possibly making the AI a goal over ther next several patches. Let me try this...

If you do (even once), you may be surprised at the difference it makes. Pleasantly surprised (I know: I worked in software development for a long time).

The other thing it would tell you is whether the AI changes are actually helping, hurting, or neutral.

For example, staying with the naval changes that caused AI GB to be better at blockading ports. Everybody agrees (I think) that this made GB a little tougher to beat. Not much, but a little.

Let's go one step further: Say, in next change, you figure a way to optomize the numbers needed to blockade a port. For instance, if France has 10 ships in port, how many ships should blockade them? Let's say 15. Can these kinds of numbers be supported? If so, then let's shoot for that ratio in EACH port.

Then, see how GB fares. I guarantee you humans playing France will have more difficulty breaking out of port. If you come up with the right ratio, you may even keep them in port for GOOD. Imagine that. Two small changes, and France and GB stay at war for much longer.

The players, however, will see the new ratio and let you know whether it's working or not. Let's say you aimed for 1:1 instead of 1.5:1. This would probably see France try to break out a couple of times, and he will succeed on about 1 in 6 of them. Oops, London falls. Next release, try 1.2:1. Whatever.

Incremental changes carry a MUCH higher probability of working correctly than massive changes. Plus, you can find out whether they are worthwhile or not much more quickly than with big changes.

[Jimmer]

ORIGINAL: NeverMan

I just want to say that if you are the only developer on this game then I feel sorry for you Marshall and shame on Matrix. The game is so buggy and messed up right now that there should be a team working on this to fix this game ASAP, at LEAST to get it to a credible level. Right now, it's not. I imagine that it will probably take 1.09xx before it gets there.

Smaller patches released more frequently (which it seems is what you did with 1.02x) would be more desirable. It would be better for Matrix's reputation and for the players who paid for the game and easier on the testers and probably easier on you.

Exactly my point (or, one of them, anyhow): Matrix really LOOKS GOOD if they are making smaller but more stable changes. 1.02 tried to do this, but there were so many big things in it that it failed.

Let's take a for-instance: About halfway through the beta-testing phase, Marshall perfected the code for whatever that feature was he was working on. So, he added it. But, that cost at least 5 letters to go by before release.

A "drop-dead date" would mean that would have had to wait for 1.03.

Let's take a vote: Take your pick:

1) Of your 10 "favorite" bugs to fix, would you rather have only 1 fixed, but fixed rock-solid and in a month, or

2) All 10 get fixed at once, but it takes 3 months to release and causes other problems in the process? Plus, 3-5 of the 10 don't actually get completely fixed.

Those are ratios from my mainframe development days. The time spans were longer (it was about 6 months between releases -- but there were 200 developers working on them). But, the "half of the 'fixes' fail" is pretty much right on target if we tried to put too much into a release. It eventually required having a manager whose pay package was NOT dependent upon "making release" to rein in development.

Face it: We programmers always want perfection. We want all the bells and whistles, too. But, there are basic market realities that prevent the engineer mentality from "winning". Managers must step in and tell the programmer to stop writing features and start earning money: By RELEASING!

[pzgndr]

Incremental changes carry a MUCH higher probability of working correctly than massive changes. Plus, you can find out whether they are worthwhile or not much more quickly than with big changes.

At this point, a couple of comprehensive updates for v1.03 and v1.04 make sense. The bug list has lots of game bugs and rules deviations which need to be worked off. This is tedious but should be straightforward. Most of the AI work for v1.03 is necessary to get several fundamental behaviors working better, and in turn should get more people playing again who can then provide more feedback. Personally, I've just been poking around with it for the past month or two because I'm not learning much with the current AI and concerned I may be learning wrong lessons. A few modest improvements with the DOWs, alliances, stacked attacks, unit production, etc. should help a lot. I'm guessing other newbie players are in the same boat.

Once those things are resolved, then incremental updates to introduce game enhancements, UI enhancements and long-term AI enhancements should be considered. Marshall being one guy, I'm doubtful a monthly update schedule for official updates is practical. Better to identify a series of feature updates for an upcoming release and then post betas every week or so as things get added? Whatever, I expect we're in for a long haul over the next several months to work off everything on the current bug list.

Alternatively, we could call J.G. Wentworth and get the final version now? IT'S MY GAME AND I WANT IT NOW!! [:D]

[MarshallEllis]

ORIGINAL: NeverMan

I just want to say that if you are the only developer on this game then I feel sorry for you Marshall and shame on Matrix. The game is so buggy and messed up right now that there should be a team working on this to fix this game ASAP, at LEAST to get it to a credible level. Right now, it's not. I imagine that it will probably take 1.09xx before it gets there.

Smaller patches released more frequently (which it seems is what you did with 1.02x) would be more desirable. It would be better for Matrix's reputation and for the players who paid for the game and easier on the testers and probably easier on you.

Neverman:

I have always been the only developer. This was no secret. It's tough I know (to tackle a monster like this with one coder) but nobody held a gun to my head. I knew this was going to be hard. We have made progress but I also know that we have more to do!
I appreciate your candor.

[baboune]

Marshall, is there some plan for Matrix to have an EiA 2.0 release somewhere in the future once EiANW is satisfyingly patched?  And if so is this why you are still working on this?

[MarshallEllis]

Actually there will be WAY more than 2.0. Our engine has tanks, airplanes and much more! EiANW was just the first wagon across the river and the first wagon takes the most arrows ! I would love to do more with the EiA portion. I would love to do a classic EiA with original counters and maps but really you guys will help determine this.

[NeverMan]

ORIGINAL: Marshall Ellis

Actually there will be WAY more than 2.0. Our engine has tanks, airplanes and much more! EiANW was just the first wagon across the river and the first wagon takes the most arrows ! I would love to do more with the EiA portion. I would love to do a classic EiA with original counters and maps but really you guys will help determine this.

If you came out with a pure EiA original "classic" game that was fairly bug free at release, I think most here would dump this buggy hybrid and jump on the classic game. JMO. I know I would.

[Soapy Frog]

Yes. An EiA that encompasses the original rulebook is all I could possibly desire.

[delatbabel]

A scenario editor I think would be the best bet, then we could recreate the original EiA game as well as the later and earlier (1792, 1788 etc) scenarios.

[pzgndr]

Actually there will be WAY more than 2.0. Our engine has tanks, airplanes and much more!

This would allow us to answer the burning question, "What if Napoleon had a B-52 at the Battle of Waterloo?" [:'(]

[Jimmer]

Well, I for one, don't find ALL of the EiH rules bad. The map, for example is much better, at least on land. Plus, the idea of having every neutral minor gain factors when DOWed, so they're not pushovers, that's good, too.
 
Plus, there's no way we'll EVER get pure EIA as it was in the box (with or without errata). Can't be done, because of the programming constraints. Even TCP/IP wouldn't allow for it. It's just too complex. For example, there were about 20 steps to just the diplomacy phase. Something had to be given up in order to make a playable game.
 
But, making all potential changes be optional would be an excellent compromise. Until we have to draft a 20-page contract specifying which rules we're playing with ... [;)]