Solinear,
That''s an interesting way of looking at it
The client''s computer could bear some of the ai burden, however. Up to a cetain point, it may even be able to take care of all of the ai that that player would encounter.

For example, because of the large map, and the diversity of the people playing on it, meaning timezones, you wouldn''t meet too many new people at once; unless you happened to stumble upon a huge battle. Now lets say that the server has ever player''s ai scripts logged, and stored. As soon as a player comes in range of someone who is offline, now this happens long before the player is aware he is about to meet someone, the server send the ai script to the client. As soon as the player actually meets the other race, the ai script starts running, but on the client''s computer. Now this might be done for 4,5 various ai, but most likely no more per client. Considering the magnitude of the game, you aren''t that likely to meet multiple offline users anyways.

A final thing I want to add, is that you won''t necessarily know the user is offline. There will not be a flag above their heads reading: I am offline, come and kill my ai Players would proceed upon the assumption that they are facing a human opponent, and continue from there.

One more thing I forgot, there are lots of aspects of this game other then combat; although combat is a major focus. Trade, diplomacy, economic are all other things that we have to think about when the user goes offline.

Thank you all for comments, and I''d like to know what you think.

Thanks again,
Fireball

quote: For example, because of the large map, and the diversity of the people playing on it, meaning timezones, you wouldn''t meet too many new people at once; unless you happened to stumble upon a huge battle. Now lets say that the server has ever player''s ai scripts logged, and stored. As soon as a player comes in range of someone who is offline, now this happens long before the player is aware he is about to meet someone, the server send the ai script to the client. As soon as the player actually meets the other race, the ai script starts running, but on the client''s computer. Now this might be done for 4,5 various ai, but most likely no more per client. Considering the magnitude of the game, you aren''t that likely to meet multiple offline users anyways.

This is IHMO not a good idea. If you use the client to perform AI, you will open up the door for cheats. The server could randomly check the validity of some actions, but it would still leave some holes.

Also if it wouldn''t be likely for you to meet offline users, then it would be even more unlikely to meet online user. Either this would become a boring game or this view might be wrong.

Greetz Safariman

First off, cheats are a problem, I agree; we will worry about them later. As for more people being offline then online, this is simply not true. As in the case of any mmo, people being offline is always a problem, but the game has to deal with it. I think that with some tighter security, running ai scripts of the client won''t be too big of a problem.

quote: Original post by fire_ball
As for more people being offline then online, this is simply not true. As in the case of any mmo, people being offline is always a problem, but the game has to deal with it. I think that with some tighter security, running ai scripts of the client won''t be too big of a problem.

Actually the highest player usage that I ever saw on Everquest before they pulled the user counts was 65k and that was when they had just over 300k user accounts. The only times that they ever went over 65k was on Sundays, when they hit 72k for a very short period of time. In other words, the most addictive and prolific MMOG in the US never hit 25% and didn''t get over 20% for more than 15% of the time.

AI on the client is flat out a bad idea though. As someone said in the past: "never, Never, NEVER trust the client. He is the enemy". That might not be an exact quote.

Tight security doesn''t ever protect you. They will crack your security. I''m sure that you''ve got the best security people in the industry, but it''s still not going to protect you against packet sniffers and packet disassembly.

I think that my advice is going on deaf ears, but I still think that it''s necessary to give it.

The TradeWars 2002 idea is a good. I too remember that game and the good thing about it is that there were certain protected zones where you could leave your ship and could not be killed when you left the game. So you can leave your ship there instead of on your planet which if not well defended could be destroyed. But the catch was that only a certain number of ships could occupy each protected zone (I think there were like 6 protected zones) so at midnight excess ships whould be kicked out of the protected zone (the game kept track of who was there first) into some random place in space.

I would like to say that client-side uh... stuff, scripted... whatever you were talking about is just fine. Infantry did it, they had all of the numbers(most of them) on the client side. Everytime you started the game, it checked the document with the numbers, checked for an update. I think it did this by size and a couple codes. Changing anything meant adjusting the size and the server not letting you play, even if you did change that, the codes would have to match up with certain things. Basically I think security is an easy issue to people who spend a little thinking on it, not to down-play anyone who has had problems. I really don''t like cheaters either... they basically ruined diablo, and part of starcraft(didn''t play wc that much). I know you don''t wanna be bogged down with checks, but if you hide, bluff, and cover... you should be fine.

"Practice means good, Perfect Practice means Perfect"

I just wanted to react to someone (forgot who) that mentioned the raised defense AI and stuff:

I suggested this on the project forum too, there shouldn't be a better defense when the player is offline, just more focus on defense. (The player can do this while he/she is online too, but this wouldn't lead anywhere) The defense value should always be high (not only during offline, always) because you wouldn't want your empire to be destroyed in one attack (or ten for that matter). This should take time and effort.

About the 'new players':

If I recall correctly, new players will be put in a 'new' area (with other new players), where they will have grown by the time the settled players reach them.

----------
Visit http://www.endoentertainment.tk, http://www.halfdemon.tk and http://www.thedarklands.tk

[edited by - Half-Demon on August 4, 2002 8:03:54 AM]

Yes, I agree with Half-Demon, who is in fact a designer on our team When logging off, the player''s ai should not completely go into defensive mode, but you can''t really expect it to expand the empire much, or take too many risks. As for cheating, I agree that someone who wants to prevent it and thinks about it can do so without much trouble. But any online game runs the risk of being hacked, regardless. You just have to deal with it in real-time.

Thank,
Fireball

fire-ball, dealing with cheating is not something best dealt with on a "as it happens" basis. you should always be thinking of possible exploits and trying to prevent them as early as possible. the hardder it is to create the first few cheats, the better off you are. hiding something that was once known is quite difficult since now you know what to look for and what it will look like once its found.

KingRuss has the exact oppisite view. ist NOT easy to just prevent cheats. the ppl who create them are quite resourceful, and have the entire source code to deal with. they can make sure that changed client side stuff wont get noticed by the server. this is why you need to do as much server side as possible and not give the client any more info then it needs.

remeber, the actually ppl who create cheats are not bad. normally they do it for fun, and are quite smart. its the morons who download them off the net and use them that are bad. a few cheaters are fine, since they wont ruin the game for the majority. in fact you want to ensure that you dont boast about your security, it will make the cheating seem so much more fun. though having security is a definite.

solinear is correct, however ppl creating cheats dont need packet sniffers at all. they have access to the entire client executable. they can use pass through dlls, modify things in the games memory, read from the memory, change files when loaded into memory, map windows calls to their own, etc. for instance, hl with all its spyware cheat protection is stil not immune. sure popular cheats get detected quickly, but someone could use dlls and wrap the entire OS and thus the game would never be able to deatect a thing. sure such a cheat would be difficult to code, difficult to install, and likly error prone if the coder is not careful. i mean theoretically you could emulate the cpu thus the game cant even trust the processor its running on. i agree with you, i think he dont care much about hearing things other then he wants to in regards to security.

some things you cant worry about, though its not an excuse to disregard all exploits and have little protection.

I agree, one thing I had suggested earlier, was that (because people can be very resourceful) ''we'' should try to think of every possible action a player could undertake (I said ''try'' ) to limit the chance of them finding a way to do things they aren''t supposed to do.

----------
Visit http://www.endoentertainment.tk, http://www.halfdemon.tk and http://www.thedarklands.tk