Combi VS Purri

[console]

Yes but the client can tell the server, I fired and I fired AT the opponent and no where else.

Sort of... This is how the server reads movement and button information from the client:

Code: [Select]

---quake2/qcommon/common.c---
void MSG_ReadDeltaUsercmd (sizebuf_t *msg_read, usercmd_t *from, usercmd_t *move)
{
    int bits;

    memcpy (move, from, sizeof(*move));

    bits = MSG_ReadByte (msg_read);
       
// read current angles
    if (bits & CM_ANGLE1)
        move->angles[0] = MSG_ReadShort (msg_read);
    if (bits & CM_ANGLE2)
        move->angles[1] = MSG_ReadShort (msg_read);
    if (bits & CM_ANGLE3)
        move->angles[2] = MSG_ReadShort (msg_read);
       
// read movement
    if (bits & CM_FORWARD)
        move->forwardmove = MSG_ReadShort (msg_read);
    if (bits & CM_SIDE)
        move->sidemove = MSG_ReadShort (msg_read);
    if (bits & CM_UP)
        move->upmove = MSG_ReadShort (msg_read);
   
// read buttons
    if (bits & CM_BUTTONS)
        move->buttons = MSG_ReadByte (msg_read);

    if (bits & CM_IMPULSE)
        move->impulse = MSG_ReadByte (msg_read);

// read time to run command
    move->msec = MSG_ReadByte (msg_read);

// read the light level
    move->lightlevel = MSG_ReadByte (msg_read);
}
So you can't, to my understanding, tell the server "i shot at this target".  The best you can do is say, I turned to face this angle, and I had the fire button down.

I.e. you can't just shoot at something, you can only turn, and the server will shoot at whatever your crosshairs are on.

Anyway that's my understanding from reading the code.

Regards,

quadz

[naymlis]

Is it possible to narrow the spread of chaingun bullets?

lol no

[Art]

I don't know whether to

[haunted]

bot gg

[Art]

We've already discussed that, I'm refering to the chaingun bullet thing .... kinda reinforces one of my earlier statements      

[QKmOrPh]

haunted is a freaken noobie, he calls so many people hackers it's almost like a sickness he has.    e|haunted

[naymlis]

just cause he accused you of cheating doesn't mean he calls everyone a hacker.. if you're lookin for someone like that go talk with dinsum or b100d.

[[BTF]Gator]

that demo is legit.  Nothing but humans.  No aim assistance.

[R. Borlax]

Can you kill the walls in q2 now? combi sure tried a few times.

[us/3r]

From seeing those demo's unless there was a lot of time in between them then I'd have to say they're 2 different people.  The main queues are the movement and positioning.  In the first (suspect) demo it's incredible strategy with little wasted movement when he's actually playing around.  But halfway through he just starts messing around (like 360 machinegunning above cg).  The rest of the match is characterized by twitchy movement (which is odd...maybe for speed?) and accurate and smooth aim.  If you look at the single fire weapon aiming it's dead on....rockets and SSG alike, but the second after he stops firing he goes back to twitchy all over the place movement.

The second demo his movement is clumsy and not well thought out.  He seems to have no clear strategy and little idea of where his opponent is in the level.  However if you notice even for such clumsy play his aim surprisingly accurate.  I'd say either he had no respect for the second opponent, there is a huge experience lap between the 2 demo's, or there is the possibilty of a bot....but I doubt the 3rd.

Despite the fact that it's been labeled as a cheat demo, every shot he made IS possible.  I think the throw is that his movement with his mouse is twitchy and quick, but when he aims with a refire weapon (cg) it goes really smooth and accurate, so from his movement you wouldn't be able to tell that his actual firing aim is any good.  This is also the reason I would tend to rule out botting.  The ratbots I've seen actually continue to lock onto the corpse after it's dead, and after firing has stopped.  It's a smoother lockon than say a z-bot, but it won't just lose the target mid-firing.....it's possible he turns it on and off while firing, but if you're using a low enough fov to hide the bot then he'll continually lose the target, which is where you get the twitchy locking on.  His aim is smooth and accurate, but he does miss to the left and right quite a bit even while being on the target.

My main argument is from the way the cg works.  It fires in cycles.  It has a warm-up, a cool-down, and a mid-firing cycle.  Each has it's own damage per second, and they fire in a cycle.  If you fire a cg into a wall you'll notice a kind of circle of bullets as you fire.  The bots lock onto the center of the target while the game itself figures out where the circular pattern hits.  But since the cg is actually a cycle, that means it has to do a check on the +attack toggle for each cycle, and the bot renews the lock-on for each individual cycle, so what you end up seeing if you're quick is the entire circle of chain bullets hitting the target and sort of sawing them.

A secondary argument is one that's already been brought up.  Purri was saying that he was lagging in close combat.  Once he started lagging then his reaction time for dodging slows way down, thus making his movement a lot easier for a lpb to predict.  This happens I'm guessing mostly with the cg.  I know because it's happened to me.  By the time your packets catch up you're already meat.  This can give the illusion of damn near perfect aim for your opponent, but to them you're just kinda standing in one spot. 

As a minor point also notice in the first demo that combi uses his movement keys to aim as well as his mouse.  He strafes perfectly in time with purri to keep his chain on him.  He doesn't create an impossible angle, he keeps him right in his crosshair without having to follow with this mouse. 

IMO if he's cheating there's no way to tell without an analysis of the actual packets from the match server-side.  In other words I've seen players who would regularly encounter bots and would end up with better percentage aim.  It's amazing but true.  In this case these players are both incredible, and could both easily be mistaken for bots by inexperienced bot hunters, so I say not guilty by lack of evidence beyond opinion.

And don't even try to come back with that "well they all say it's a bot, and they know".  Who the hell are "they"?  And how many bots have they seen?

[the Crazed]

Nice post user, there's some information in it.

My main argument is from the way the cg works. It fires in cycles. It has a warm-up, a cool-down, and a mid-firing cycle. Each has it's own damage per second, and they fire in a cycle. If you fire a cg into a wall you'll notice a kind of circle of bullets as you fire. The bots lock onto the center of the target while the game itself figures out where the circular pattern hits. But since the cg is actually a cycle, that means it has to do a check on the +attack toggle for each cycle, and the bot renews the lock-on for each individual cycle, so what you end up seeing if you're quick is the entire circle of chain bullets hitting the target and sort of sawing them.

To clarify, the circle of bullets saws across the center of the target as the bot twitches to renew its lock-on through the cg cycles?  So you would only notice that happening at the time the cycle is switching, right?

 

The rest of the match is characterized by twitchy movement (which is odd...maybe for speed?) and accurate and smooth aim. If you look at the single fire weapon aiming it's dead on....rockets and SSG alike, but the second after he stops firing he goes back to twitchy all over the place movement.

I think that if you were to take a look at demos of the top echelon of players you would find they're ALL very twitchy movers.  If you quickly twitch the mouse you can position your crosshair to circle/strafe jump at the highest speed - and you can do it midair too if you find your aim is getting too far off from your direction of movement (or before it gets to that point - which is usually the case).  For example, if you're doing a right strafe/right circle jump and your aim drifts so far to the right that you'll start to slow yourself down you can release the movement keys (in midair) twitch the mouse left and then resume your right circle mouse motion while right-strafing to complete your jump.  I think that's how it works, someone correct me if I'm wrong.  Purri is twitchy as hell, and it doesn't seem to prevent his aim being smooth when it needs to be.  Think I've seen a demo of him circle/strafe WALKING to squeeze out the last drop of speed.  Crazy.  I'm starting to see that twitchiness as an indicator of skill.

Anyways, if there's no visual indication... er... PROOF... that he's botting - then do we shift to accuracy information to determine a bot then?  Somehow that doesn't seem fair but then again.... hard stuff to judge.

[us/3r]

My main argument is from the way the cg works. It fires in cycles. It has a warm-up, a cool-down, and a mid-firing cycle. Each has it's own damage per second, and they fire in a cycle. If you fire a cg into a wall you'll notice a kind of circle of bullets as you fire. The bots lock onto the center of the target while the game itself figures out where the circular pattern hits. But since the cg is actually a cycle, that means it has to do a check on the +attack toggle for each cycle, and the bot renews the lock-on for each individual cycle, so what you end up seeing if you're quick is the entire circle of chain bullets hitting the target and sort of sawing them.

To clarify, the circle of bullets saws across the center of the target as the bot twitches to renew its lock-on through the cg cycles?  So you would only notice that happening at the time the cycle is switching, right?

Exactly. Except with the better bots I've seen there's more of a scanning feature which is almost indistinguishable from a human player, unless you follow them perfectly through a 360 ellipse pattern.  As you'll notice from the first demo though purri never manages to get behind combi, so there's no way to tell.  Even with a low fov the bot will maintain the lock past 180 degrees, but even that could be done by a human who uses a really high sensitivity and is good with it, or someone using a trackball.

 Usually what you'll see in an unlagged opponent when they start getting hit is strafing left or right to avoid the CG and create an angle that someone would have to "reset" they're wrist to maintain, then move the other way and that's what causes the opponent to miss.  But the bot will continue through an impossible wrist angle through every small change in position.  Ratbot is notoriously smooth in this, but it's unmistakeable after a certain point.  Think of a swoosh (like a Nike insigina) toward your opponent.  As you get closer it's damn near impossible for them to have the same accuracy as if you were just moving left or right.  If you don't move far enough though there's no angle, and if your opponent is savvy they can strafe with you, which is what combi does very well.  Does that prove he's not botting?  No.  But if you can predict that well there's almost no point to using an aim-assisting program, imo. 

Anyways, if there's no visual indication... er... PROOF... that he's botting - then do we shift to accuracy information to determine a bot then?  Somehow that doesn't seem fair but then again.... hard stuff to judge.

That's my point exactly.  Accuracy information can be misleading, because people can have really good games where they're just in their opponents head, and games where they can't hit the broadside of a barn.  If purri was off and combi was on at that level there's little doubt it would appear one of them was cheating somehow.  From demo's I've watched of naymlis you could almost say the same about him and his flick rails, but really it's just from much practice and muscle memory. 

I think the biggest giveaway really is that purri didn't notice until someone showed him later.  Usually if you've played against someone before you know when they're having a good game, and when their aim is being corrected.  Obviously combi and purri had some history in which combi chained a lot, so it wasn't a huge shock to purri who kept getting lagged when in close.

-note: my opinion is not endorsed by myself or anyone else.  Any similarity between my opinion and anyone else's is purely coincidental, and therefore a product of llama's.