For the playtesters: How is the AI?

gambler · Post by **gambler** » Wed Oct 15, 2003 9:41 pm

Greetings:

I realize I'm unlikely to get a wholly unbiased answer to this question since folks who respond will probably be trying to get me to buy the game.

However, it's unlikely that I'll have a lot of time to spend on head to head play, so AI is a big deal to me. I have to say, I was somewhat disappointed reading some AARs since it seems like in most of them the player achieved decisive victory, sometimes even after increasing the handicap.

Yes, I know, AI will never be as good as a human player. But I had high hopes because I (as someone who's never designed AI), assumed that a plan oriented AI would be easier to create than an AI that had to do both plans and determine that to complete them two 4-1 attacks here, followed by movement of these three units to set up another attack would be the best way to do it. I don't mean to sound negative, but if the game is one in which a player can expect a decisive victory using sound play it won't hold my interest for very long.

MarkShot · Post by **MarkShot** » Wed Oct 15, 2003 11:21 pm

I am a beta tester. However, I don't want to influence you or anyone else to purchase the game. I am just sharing my perceptions and you and others can make your own decisions.

In regards to the AI, these are my preceptions:

(1) The AI gives you a run for your money. In playing/testing for almost 1.5 years, I haven't found myself grumbling how stupid various actions it has taken are. This is not one of those games where the AI is only good for some learning curve action as a target drone and you must find online opponents for any fun. I've been playing SP exclusively and have been having a great time.

(2) The AI is still computer software and after enough play it will be possible to pick up certain patterns. It's not going to learn, evolve, or pull off amazing and radical tactics. The actions it generates are reasonable, but the more familiar you get with it, the greater the chance to exploit that understanding. Of course, the same is true when one regularly plays online with certain individuals too.

(3) You commented on AARs demonstrating that sound strategies lead to victories. Well, that seems appropriate, since the measure of a simulation is when what works in the real world also succeeds within the simulated model.

(4) You commented that most or all the AARs posted demonstrate victories. I tend to think in gaming communities that is generally true. I guess human ego runs against players putting up "My Web Page of my Greatest RDOA Defeats". The other reality is there is a fairly limited set of AARs out there. Many have been written by Yakstock who is probably one of the top players out there. The few others have been from testers with lots of war/strategy gaming experience and intricately familiar with the game. I am personally amazed sometimes how many losses I suffered when starting out and didn't hang it up. Of course, I admit to being clueless when I started. Still, I have taken a number of losses over the last six months.

(5) I did actually post a loss on a thread yesterday. And I posted a marginal victory on another thread a few weeks ago.

(6) You refer to AAR victories despite the playing giving the AI some bonuses. I always played the defaults. I would be surprised if I could still win having altered supply and reinforcements to favor the AI. In regards to playing with order delays enabled or disabled, it is not clear to me whether this particularly favors either side.

(7) One thing that the AI does very well is probing for weakness and finding holes in defense. I tend to be pretty impressed by it. Of course, this capability it has can be exploited, since it may tend to probe and shift instead of mass. The reality is that whenever one combatant can see into the future, he or she has an inherent advantage whether it be against the AI or a human.

(8) One thing that I think I do better than the AI is move my arty units and keep them safe.

(9) The AI does not cheat. Meaning it follows the same rules you do. Any prowess it demonstrates comes from the quality of the software as opposed to short circuiting the rules that the player must follow.

So, there you have some of my perceptions. I am confident that the AI is the best you can get on the current class of hardware and for the price of computer game in 2003/2004.

I hope that helped.

JeF · Post by **JeF** » Thu Oct 16, 2003 11:38 am

I'm no Beta Tester and can only speak from RDOA experience.

Markshot expressed a very good view of the AA AI. Maybe his last sentence about the "best AI" is a bit too much. But it is a tough opponent. And, not only in prepared defense. Beware of flanking counter-attacks !

You'll have fun. I lost so often (not that I'm any good :rolleyes: ) with the BF.C demo that, for a time, I was unsure to buy the game or not. Finally I bought it and don't regret any cent. A marginal victory is still a victory to me.

BTW, Yakstock victories do not count, as he is certainly the best AA player out there.

Cheers,

JeF.

Grouchy · Post by **Grouchy** » Thu Oct 16, 2003 6:40 pm

I like the AI in HttR.
Most AI's can defend pretty fine, but a lot of them have problems attacking.
Still remember downloading the RDOA beta and the really pleasant surprise I got when I noticed the AI. Finally an AI that could attack!

I admit suffered some defeats in the Arnhem historical campaign as the allies, and can tell you from my own experience that if you give your para's the same orders as where historically given on the first day (don't know if this is sound play, but hee....those officers in '44 thought it was a sound plan

), your para's will never reach at all or in so small numbers, that they won't be able to hold the bridge for long.

gambler · Post by **gambler** » Fri Oct 17, 2003 9:38 pm

Thanks for the replies all, especially MarkShot's lengthy one. I'm still somewhat debating although I'm leaning more towards my usual "Wait for the first good review of finished game to convince me it's a good buy." mode.

Yakstock · Post by **Yakstock** » Sat Oct 18, 2003 2:11 am

Gambler,

I wanted to weigh in since you are still "on the fence". My honest opinion on AI is that it is as good as any other I have played against. I would give it an 8 out of 10. I realize you were focused on the AI, but I want to add my thoughts on the human versus human aspect of this game. In this category I would without a doubt classify this as the greatest game I have ever played.

If you really want to have fun again and again, get into human versus human play over the Internet. Every game will be different and I guarantee it will capture your attention for a long time. I just finished a game of the old AA: RDOA that lasted 7-months (we played at slowest speed and only caught up mostly on the weekends). It was great fun.

If you think you will get into this type of play, then get off that fence and come on down to the best game you will ever play. If you will only play AI, then perhaps you can sit on the fence for a while....

Yakstock

PS - I appreciate the comments on my play in this thread, but I am really only lucky and many are as good or better than me.

MarkShot · Post by **MarkShot** » Sat Oct 18, 2003 2:15 am

Gary,

Who was it that said, "I would rather be lucky than good any day."?

I don't recall.

Yakstock · Post by **Yakstock** » Sat Oct 18, 2003 2:48 am

Lefty Gomez, Hall of fame pitcher I believe - and I agree....

Kevinugly · Post by **Kevinugly** » Sat Oct 18, 2003 7:28 pm

Gambler, because of it's nature, AA stands or falls by the quality of AI more than any other game in the genre. I played RDOA practically to death when I bought it and probably found every hole in the AI there was to find. That said, I would still put the AI up with the very best and am looking forward to dissecting HTTR when it comes out. It's not quite as challenging as playing another human, having just lost to Yakstock in that 7 month epic (I'll get you next time :p ) I can't go back to RDOA's AI as it's just too easy

MarkShot · Post by **MarkShot** » Sun Oct 26, 2003 3:56 am

Gambler,

There you go. I just completed a 5 day brutal fight as the Germans in the Nijmegen - Mook Hook Right scenario. The options taken were all the defaults (meaning I probably would have lost if I gave the Allies any bonuses or myself and any handicaps) and ORDER DELAYS = PAINFUL.

With all my skill of 1.5 years at this game, I fought the AI to a draw. So, no, playing against the AI is not a piece of cake.

PS: I am using the RDOA tiles to play (my preference) and that's why it looks different from the other HTTR screenshots which you have seen.

Beery · Post by **Beery** » Thu Oct 30, 2003 3:08 am

I would imagine that the fact that you are required to use AI as a part of your own command structure would level the playing field - in other words it will make the enemy AI appear to play that much better. Games which allow individual control of units allow players to exploit weaknesses in AI, but as I understand it, this game would prevent players from doing so because your orders must be interpreted by your own AI subordinate commander. That's how it worked with the Battleground series 'commander control' feature, so I assume it will be similar with this game.

MarkShot · Post by **MarkShot** » Thu Oct 30, 2003 3:24 am

You could actually give orders to every unit on the map. However, it would be counter productive to do so. Just like if you are company commander which means that you should be the most senior/experienced officer in the company. Therefore, you could give individual tailored orders to each soldier in the company, but the results would be suboptimal.

In game terms, you would pay the following penalties by giving orders to each and every unit:

(1) Excessive order delays.

(2) Lack of cohesive action by units on defense and a total inability to even contemplate making an attack.

(3) In ability to manage the big picture which is the role you have been cast in as the human player.

...

At other end of the spectrum could find the senior commander on the map and order him (meaning the entire force on the map) to simply attack here or defend here:

(1) Of course, this doesn't work if there are multiple objectives.

(2) Does not allow you to determine which forces are used best in which roles.

(3) Does not allow you to take maximum advantage of different terrain features of the battle space.

(4) Does not allow you to meter the level of casualties/fatigue you sustain so that you can best pace yourself.

(5) Does not allow you to attack along different axis or defend at various locations or layers.

(6) Does not allow you to formulate a plan that uses time as a strategic dimension.

...

So, what's my point? After all, I should have one.

Simply this ... playing (commanding) HTTR effectively is finding the right balance between micro/macro managing. This is partly a function of the dynamics of what works well in the game and partly a function of one's individual style of play. More so than any other game which I have seen, this game allows the maximum flexibility of finding a balance somewhere along that spectrum which I enumerated above. Also, HTTR allows you to within even one gaming session micro-manage one area of the battle while you macro-manage another area of the battle. It really is quite unique.

I am sorry, but I am not familiar with the game you mentioned, but I hope my explanation helped you to understand it a little better. Also, see:

showthread.php?t=44457

Golf33 · Post by **Golf33** » Thu Oct 30, 2003 3:49 am

For players who like to micro-manage, the order delays are optional. This removes the response-time penalty that comes with micromanagement with order delays enabled.

Personally I don't like to micro-manage except for small scenarios and assets that are normally commanded at a high level anyway, like AT and artillery. I do like the manoeuvre aspect that comes with orders delays so I keep these set at the maximum.

Regards
33

MarkShot · Post by **MarkShot** » Thu Oct 30, 2003 3:52 am

Yes, I did forget to mention that those could be turned off.

I've gotten so adjusted to order delays that when my wife asks me to do something and then asks an hour later why haven't I done it yet, I just say, "Order delays = Painfully realistic".

Arjuna · Post by **Arjuna** » Thu Oct 30, 2003 12:05 pm

MarkShot wrote:I've gotten so adjusted to order delays that when my wife asks me to do something and then asks an hour later why haven't I done it yet, I just say, "Order delays = Painfully realistic".

gambler · Post by **gambler** » Thu Oct 30, 2003 7:29 pm

And then she responds with, "No, just painful for you..."?

Seriously, thanks for the reply, this game has been preordered by me so I'll get to find out how strong the AI is first hand. Given I'm getting my rear handed to me in games like Galactic Civilizations and Starships Unlimited (admittedly different beasts), AI doesn't have to be spectacular to do a good job against me.

MarkShot · Post by **MarkShot** » Thu Oct 30, 2003 8:48 pm

Gambler,

If you would, when you get the game, let me know how well or poor my tutorial had conveyed the style and feel of play for HTTR. (and, of course, some details)

One my goals in writing it had been to bring the game to life for those who had not owned RDOA. I am curious to know if I succeeded.

Thanks.

MarkShot · Post by **MarkShot** » Sat Nov 01, 2003 4:25 am

Maybe we should retitled this thread to "MarkShot's humiliation at the at hands of an non-sentient AI

".

Well, I have done it again. I just played a 2 1/2 day scenario titled "Crunch Time". It is essentially a meeting engagement with four objectives.

Now, prior to the battle as I usually do:

(1) I had analyzed the terrain.

(2) Formulated a well thought out plan of phases with a time table.

(3) Issued clear and detailed orders reflecting my plan.

Now, for the first 25% of the scenario, things seemed to going the way my plan would have predicted. But like someone had said on this thread, the AI did a very credible job of attacking the two objectives which I had made my Phase I/II objectives while holding on to the other two objectives. After securing them and digging in, I had planned in Phase III/IV, I was going to swing around with my armor and take the two other more exposed objectives.

Man, I really had a solid plan and if I was to detail here if I didn't say anything more about the how the game went, you would have been suitably impressed. Well, by the 50% mark, it was clear that things were not going according to plan. If in the final 25% of the battle, I hadn't managed to disengage some armor and strike a distant weakly defended objective, I would have lost. As is, I just pulled out a draw.

There it is, the AI against the pompous and presumptuous Beta Tester who drawls on and on in a "Tips and Tactics" thread.

Arjuna · Post by **Arjuna** » Sat Nov 01, 2003 4:50 am

[QUOTE=MarkShot]Maybe we should retitled this thread to "MarkShot's humiliation at the at hands of an non-sentient AI

".[QUOTE]

Mark,

Humility is always a virtue worth cultivating.

Dave

gambler · Post by **gambler** » Sat Nov 01, 2003 9:29 am

Markshot: I'll be sure to let you know my impressions. I can say that your tutorial/AAR had a great deal to do with my choice to pre-order the game. Generally if an AAR is more about the 'story' of the battle than the mechanics, it tilts the decision towards buying. Given that your AAR was also a tutorial and it still sounded enough like a story to impress me, it definitely helped my decision.