CMSF post 1.02 still worse than CM1

[sandy]

I have has enough of Steve ignoring my words, so I shall post somebody else's words to be ignored.

Here is a copy of my (anonymous) email opponent's words that came with the latest game file received this morning:

Sandy

What I find most pathetic in this implementation - and I trust it is not

irredeemable - is the complete absence of any Tac AI.

For example, on your right flank - over the last two minutes - I moved a

couple of Bradleys up and one was positioned to fire on the BMP to its front

that my infantry had seen. It moves up, but doesn't fire. Your BMP fires at

the Bradley for practically a whole minute with no apparent affect. Next

turn I have to instruct my Bradley to return fire - which very quickly

knocks out the BMP. What I didn't know was that you had another BMP on my

flank, that now has a Bradley's flank presented to it at point blank range.

It doesn't fire. Eventually my Bradley notices the BMP and blows it up.

Absolutely ridiculous. It would have been ridiculous 10 years ago, and is

unbelievable now.

What is outrageous is how sanguine Battlefront appear to be about this, and

how there appear to be a core of players who deny any such shortcomings.

It never would have happened in CM1

We have paid to beta test this game...Now I would have happily beta tested for free, and quite probably even paid to beta test if I had been asked politely and honestly. But I preordered a wargame that was supposed to be better than CM1.

[ August 17, 2007, 03:17 AM: Message edited by: sandy ]

[Merkin Muffley]

To be fair, there must be some Tac AI operating, otherwise

Eventually my Bradley notices the BMP and blows it up.

wouldn't have happened. But its apparent absence will always be more noticeable than its functioning, especially when it seems so prevalent. I know this is not news, but it also seems to show itself when the individually-modelled, graphically impressive members of a squad stoically remain in positions where they are getting shot to pieces rather than just cracking, or making a controlled withdrawal.

I am trusting that these sorts of issues - together with the more boggling aspects of the pathfinding - will be sorted in forthcoming patches. Then, at least, the game can be properly assessed.

I am sure a lot of the criticism of CM:SF and comparisons to CM1 games are that it is still work in progress. But these comparisons, whilst invidious, are inevitable while the game seems so undercooked in some critical AI areas.

[FaxisAxis]

What I find most pathetic in this implementation - and I trust it is not

irredeemable - is the complete absence of any Tac AI.

This is what I worry about the most: But is it redeemable and how much so?

[sandy]

Merkin - Of course there is some AI present.

My point is that after patch 1.02 it is much, much worse than CM1, so bad as to spoil the game experience.

So what was the point of releasing 1.00 or 1.01, and who (did anybody) beta tested them?

My opponent and I are picking up these and other problems in the early stages of our very first PBEM game...did I mention our units set up in the same place on the same side of the map?

This is supposed to be a released game that has already been patched twice. I don't think that I am such a great beta tester, I am thinking the game was not properly tested.

PS My non firing unit had its gun barrel about 10 m from the side armour of a Bradley that had not spotted it for about 45 seconds. Both vehicles were stationary. The Bradley then rotated it's turret and killed the BMP through the frontal hull armour with its second shot. That same Bradley had been hit by about 50-60 BMP shells over the preceeding two turns, fired by another BMP that it also killed with it's first burst.

My BMPs were Republican Guard in command control in a good state, not fired upon previously - ie about the best a Syrian APC can be - yet it did not shoot for near-certain kill.

I do wonder how much is bugs and how much is inherent pro-US bias built into the game.

[Merkin Muffley]

Having caught up on other threads elsewhere after posting the above I see that Steve has said pathfinding and Tac AI are being addressed in the v1.03 patch for which I am truly thankful - and many thanks to them for busting various guts to try and improve the position as fast as is practical.

I'll keep my gob shut pending v1.03 becoming available.

[Becket]

Originally posted by sandy:

So what was the point of releasing 1.00 or 1.01, and who (did anybody) beta tested them?

Steve explained in another thread that the game was released because the contract with Paradox required it to be released. They had asked for and used all their contractually permitted delays and had it been in their control, would not have released it.

[Capt. Toleran]

Let's not forget the downside of Tac AI -- tanks and other vehicles running away in crazy ways, exposing flanks, doing other things that drive us nuts. Pathfinding issues aside, I still have seen less of that than I did in CMx1. It used to drive me crazy that a tank with a good chance for a flank shot on a Tiger or something like that would consistently refuse to stand it's ground and fire, despite repeated orders. Have some faith in BFC.

[sandy]

Capt. T.

I agree it was occasionally annoying when, say T34s would reverse away instead of taking a good flank shot. But this happended overall quite rarely and the tac AI in CM1 was, in my opinion, excellent, and really made the game work well and fun to play.

The CM1 tacAI is far, far better than the equivalent which we now see in CMSF 1.02. Is it soooo stupid to expect the new improved version to be at least as good as the old obsolete dinosaur?

I know which I prefer...

[Battlefront.com]

Sandy,

I'm not ignoring anything. There are only so many posts I can make in a day. I have other things to do and I already spend 4-5 hours here.

---

People are DEFINITELY overrating the CMx1 TacAI. It did all kinds of stupid stuff, especially in early CMBO days. I know this better than anybody except for Charles... TacAI is exceptionally difficult and it will never, ever, in a million years be perfect. Yet unless it is perfect people will complain.

The difference between CMx1 and CM:SF is CMx1 had a total of something like 25 releases and 6 years of development before we stopped supporting it. CM:SF has about 3 years of development and 3 releases. Since TacAI is all about programming and not about conceptual stuff, in theory we will need about 3 more years and 22 more patches before we get CM:SF to the same standard as CMx1. Obviously we'll try to do better than that

Steve

[sandy]

Steve,

Point taken - work in progress, and all that - (must try harder and all that)

But its was great, the best TacAI despite the flaws you are aware off - I am reassured that you seem to be saying that there is no reason why the CMSF tacAI cannot be just as good if not better, given time.

I understand it can only get better, and I await great things from the TacAI by patch 1.12, or whatever, of CMSF, and more importantly, patch 1.01 of CM2WW2

One more thing...

You are right that your time is better spent working on the game than posting on the forum...but please, please tell me you will at least try to bring back some form of unit picks (at least as an option) for QBs in CM2WW2.

The current system is so unsatisfying.

I did not say we need a points system, if that's a problem, just some way to set up little tactical engagements with our own choices, as well as the computer picks we have now.

PS Do not remove the current system - it's fine and I would be asking for it, or something like it, if it was not there - sometimes I do want the AI to chose units. But sometimes the player (s) need to choose.

Thanks

Sandy

[ August 17, 2007, 09:42 AM: Message edited by: sandy ]

[thelmia]

I don't care nearly as much about TacAI as most people, as I like playing RT anyway.

Old CM AI was still an AI. I always beat it. Multiplayer was where it was at, and while the units did a lot of good things, they often did some retarded, useless thing that got them killed. Like run out of cover, not shoot what I told them, etc. I LOVE the additional control RT provides. Thanks for doing it.

BUT I do care about the lag. I can't play RT multiplayer without game- killing lag. It drives me nuts. There's an awesome game hiding in here, but I can't quite play it.

[sandy]

We could all beat the AI (usually) - that's why two player is so much more satisfying.

But the CM1 AI was (is) still the best ever, and the new game cannot function without AI at least that good.

[track]

we will need about 3 more years and 22 more patches before we get CM:SF to the same standard as CMx1

Me thinks this quite opposite that was posted earlier on that BF was going for limited subject and move from product to another in a much more quicker pace. So how long is the support period for CMx2 series games be (per game)?

[Guest]

I would hope that the lessons learned during those six years would allow Charles to get the CM:SF TacAI up to speed a bit sooner than that.

Honestly, I expect TacAI to make stupid mistakes occasionally -- and to be able to "beat" it if I play hard enough. However, not noticing a vehicle 10m away for a full minute is a problem for me. I'm looking forward to 1.03.

[flamingknives]

I have a theory that the AI in CMX1 was actually pretty bad, but it didn't seem so awful because of fundamental flaws in the game engine. Things like lack of artillery due to points, effectiveness, or lack thereof of machine guns, lack of ammunition resupply made the AI seem better than it really was.

Now that these errors have been fixed in CMX1, it could well be that the same AI cannot hack it anymore.

[thelmia]

^perhaps a better way to get steve's attention.

Eh, it's hard when your game is broken. It won't get fixed quickly but the forum posts continuously.

You need to remember the OODA loop. The forum is inside the devs loop. They can't get a patch out fast, but you can ask for one over and over in the meantime.

[birdstrike]

I agree with flamingknives,

there is no doubt the AI in Shock Force is lamentable, but it is interesting how, ever since the game was published, the AI in CM1 has reached never before known heights of intellect if we are to believe some people around here.

[Renaud]

I played the demo all-infantry as an experiment, dismounted all the US crews and conducted a traditional WW2 style infantry attack. My AT support was javelin teams with cover arcs.

The big discovery was that when giving all my teams/squads orders to go from point A to B across open ground, the teams would take strange diagonal paths at 45 degree angles to the direction of the waypoint and, AI being what it is, all the teams starting from the same general area would bunch up into the same paths. This reminded me of the computer-AI waypointing in CMx1 (infantry bunching up, no such thing as a skirmish line, vulnerable to arty, etc.).

The huge difference here is that in CMx1, infantry issued orders by a human player traveled exactly in the direction of the way point, and would only deviate if they came upon something impassable, under fire, into mines, etc. CMx2 appears to put your squads under the control of a dynamic AI pathfinding similiar to the tac AI controlling the CMx1 computer player.

This will be a big problem in a ww2 type scenario, in cmx2 currently the problem is minimized because you aren't likely to conduct this kind of dismounted skirmish line attack often.

I would vastly prefer that infantry units simply head directly to the way point without pathing by the AI unless they run into impassible terrain or come under fire, into a minefield etc (just like CMx1). The current pathing doesn't seem to serve any purpose...the squads aren't moving from cover to cover or anything like that, just following invisible geometric paths.

edit: if I had to guess, and I do, I would say this is an effect of the action point system which appears to be a grid across which infantry units must navigate point-to-point and are therefore forced to follow invisible geometric paths along these points on the way to the action point closest to the end of the waypoint. This would also explain why if you cancel a squad waypoint they are forced to scamble to the nearest 'action point', forcing a short unplanned move, possibly into a dangerous place (like the street).

Vehicles don't appear to have this issue or be impacted by 'action points' at all, I presume this is because vehicles are 1:1 single objects, while infantry units are many:1 collections. LOS and LOF calculations using the action points are forced to calculate against the 'locus' of the many:1 collection and the locus must always be fixed on the nearest action point.

[ August 17, 2007, 06:44 PM: Message edited by: Renaud ]

[Statisoris]

Oh boo, Steve has explained why the game is in the state it is many times. They are working very hard trying to fix it.

[jomni]

Stop complaining the same things all over again.

1.02 did not promise a TacAI fix.

Although BF mentioned that 1.03 will fix the TacAI and LOS bugs. So if 1.03 does not work out then start complaining again!