Matrix Games Forums

ORIGINAL: joey

ORIGINAL: Andy Mac

All I can promise on AI performance is no worse than stock (apart form more access to the AI for modders)

Not promising any more than that.

I suspect I have ran more turns of AI testing than anyone else on the team and I have seen it pull some nice moves and a few bonehead ones don't forget the added complexity of getting the AI to use all the new features.

I'm good with that. I suspect most of us are. I am just hoping to have it ready when the 90 degree days arrive in the summer. I might as well turn up the heat with WITP!

It's a little late for that. The 90 degree days of summer started in February here...and will run into December. [:D]

ORIGINAL: Shark7

ORIGINAL: joey

ORIGINAL: Andy Mac

All I can promise on AI performance is no worse than stock (apart form more access to the AI for modders)

Not promising any more than that.

I suspect I have ran more turns of AI testing than anyone else on the team and I have seen it pull some nice moves and a few bonehead ones don't forget the added complexity of getting the AI to use all the new features.

I'm good with that. I suspect most of us are. I am just hoping to have it ready when the 90 degree days arrive in the summer. I might as well turn up the heat with WITP!

It's a little late for that. The 90 degree days of summer started in February here...and will run into December. [:D]

Yeah, and this year's winter in Texas was unusally warm - both days of it.

All I can promise on AI performance is no worse than stock (apart form more access to the AI for modders)

Not promising any more than that.

I suspect I have ran more turns of AI testing than anyone else on the team and I have seen it pull some nice moves and a few bonehead ones don't forget the added complexity of getting the AI to use all the new features.

 
Of course we are all hoping for an AI no worse than stock!! But the new options available to the player will obviously challenge the AI enormously. So... have you added in a few 'helps' for the AI? Not 'cheating' as such, just 'helps'? ie. auto relocation of AK, AP, TK etc wherever they are needed, decreased fuel and supply consumption, shorter recovery times for units, increased experience gain from combat, decreased disruption from malaria zones, maybe even a few extra supply source bases, quick load (ie. single turn) for AP, AK, TK, etc, and whatever else you can think of. I DO NOT want the AI to play with loaded dice, but im not concerned as much if it needs to cut some corners in the background in order to be competitive and therefore fun.
 
Maybe you have already done lots of this sort of thing? If so, please dont tell us what is done!!
 
BTW, i would be interested in hearing others ideas on this? Agree?? Hate the idea of the AI getting any bonuses at all with a passion?? What?

ORIGINAL: Don Bowen

ORIGINAL: Shark7

ORIGINAL: joey




I'm good with that. I suspect most of us are. I am just hoping to have it ready when the 90 degree days arrive in the summer. I might as well turn up the heat with WITP!

It's a little late for that. The 90 degree days of summer started in February here...and will run into December. [:D]

Yeah, and this year's winter in Texas was unusally warm - both days of it.

And then there are those Ice Storms....

ORIGINAL: Don Bowen

ORIGINAL: Admiral Scott

How far along is AE to being completed?

75% ?

80% ?

85% ?

Just give us your best estimation please.

There is an old rule of thumb for programming projects. They reach 90% completion within a few months, then stay at 90% forever.

That's a correllary of an old adage in the architecture business: The first 99% of the work takes 99% of the time while the last 1% of the work takes 99% of the time.

ORIGINAL: jwilkerson
We've been doing integrated weekly builds for over a year and have been doing weekly Matrix Installer builds for a couple of months now - so "integrated testing" is 97% of the testing we're doing. Our only major issue is the AI - but one side benefit is that the delay in this area is allowing the other areas more time for "polishing" ... we just hope the polishing fixes more bugs than it makes!
[:D]

Just a stupid idea from a newcomer, but...

...why don't you first release a PBEM-only version, without AI requirements, with a later update specifically for the single-player AI ?

Because there are more people who play against the AI than play PBEM

Working AI is a gating issue for release we dont do anything without it.
 
To be honest its extrememly difficult to churn the numbers of turns needed without an AI to play at least one side so if we did it would have a lot more bugs.
 
I am finding them all the time in the AI v AI games I run.
 
Stupid things like is Singapore to tough to take if the AI play it historically, are there any rough spots in Japanese armament production, to few (or to many) devices for some types of LCU's, fort levels and impact on games, stacking levels on atolls.
 
Pretty much every night I leave AI v AI games running and then review the results in the morning which takes a while to go through all the reports but its a neccessary part of the de bugging process.
 
 

I also find it an intellectual challenge how can I make the AI th nastiest sneakiest opponent I can within the confines of the code we have.
 
I want to put something in that will have people go WOW the AI did that !!!! - I havent managed it yet but we keep trying. As I said no promises at this stage but we are trying to make the AI as good as we can.
 
A fact is the AI is at its worst in the big campaign because of the scale and interations with grand strategic thinking which the AI cannot do.
 
It actually does pretty well in smaller scenarios - its just few people want to play them.
 
 
 

ORIGINAL: Andy Mac

Working AI is a gating issue for release we dont do anything without it.

To be honest its extrememly difficult to churn the numbers of turns needed without an AI to play at least one side so if we did it would have a lot more bugs.

I am finding them all the time in the AI v AI games I run.

Stupid things like is Singapore to tough to take if the AI play it historically, are there any rough spots in Japanese armament production, to few (or to many) devices for some types of LCU's, fort levels and impact on games, stacking levels on atolls.

Pretty much every night I leave AI v AI games running and then review the results in the morning which takes a while to go through all the reports but its a neccessary part of the de bugging process.

I'm finding Singapore and Manila really tough nuts in the RHS scenario I'm playing right now. Singapore is being played historically, while Manila is being defended instead of Bataan. Perhaps we need to consider why most WWII operations tended to bypass urban areas. I think a large part of it was that it was hard to defend an urban area. An urban area is three-dimensional with very limited fields of fire, so that single buildings would soak up a company or more. I believe the rule of thumb on the Eastern Front was that you needed a corps to defend a Russian city adequately. So you should penalise a garrison smaller than an army in a major urban hex. Sure, the buildings would make assaults difficult, but a division couldn't defend more than a large town--basically it would have no ZoC in the same hex.

Of course IRL Manila was declared an open city and no troops were stationed there. If the AI can also prevent trrops from deploying there, that "problem" would be solved.

ORIGINAL: m10bob

Of course IRL Manila was declared an open city and no troops were stationed there. If the AI can also prevent trrops from deploying there, that "problem" would be solved.

Quezon declared it an open city because the Philippines didn't have the forces to defend it, and he wanted to spare the city. Yamashita did not intend to defend it either, given the one million civilians there, but a division-equivalent under Vice Admiral Iwabuchi was encircled and fought it out in a fortified perimeter, incidentally killing about 100,000+ civilians in the Manila Massacre.

ORIGINAL: Andy Mac

Working AI is a gating issue for release we dont do anything without it.

To be honest its extrememly difficult to churn the numbers of turns needed without an AI to play at least one side so if we did it would have a lot more bugs.

I am finding them all the time in the AI v AI games I run.

Stupid things like is Singapore to tough to take if the AI play it historically, are there any rough spots in Japanese armament production, to few (or to many) devices for some types of LCU's, fort levels and impact on games, stacking levels on atolls.

Pretty much every night I leave AI v AI games running and then review the results in the morning which takes a while to go through all the reports but its a neccessary part of the de bugging process.

So, it really was a stupid idea. Thx.



PS: I do like the smaller scenarios - I'd wish there were more of them.[:(]

Most of the scenarios in AE will be smaller ones. In fact, we have only two full-map scenarios in the initial release.

This is a really good news.  Was the list of available scenarios already disclosed somewhere ?

Only 4 or 5 separate places.

Try the thread on this AE forum titled "Scenarios?".

ORIGINAL: Terminus

Only 4 or 5 separate places.

...and carefully hidden like Easter eggs on maybe every AE sub-forum and the main page IIRC ?

'found it.[&o]

Yo Shark7 While you may be "HOT[:@][:@]" It has yet to reach 90 here!!!!!!!! It has reached 80 twiice but even that feels too warm to me. As for the AE edition coming later than expected thats fine since it is nice to be outside in the warmer weather soaking up the near 18 hours of daylight. The real secret is AE keeping me busy during the 18 hours of darkness during the winter. [:'(][:'(][:'(]