PPNL's bot

[naboo]

Hello everybody, i'm totally new to this forum and my first post ever here will be about a player i suspect is cheating :D

Ok so to cut it short, a while back i heard from some players that PPNL might be hacking, which i thought sounded unlikely, but i do also take botting seriously so i wanted to confirm my doubt and decided to record him.
What i saw caused me to question my certainty, so i made more demos, and with each one i got more sure that he was cheating.

Later it came to the point where i alerted the present admins about my suspicion, but feedback was mixed and only a few actually wanted to see a demo while some advised me not to start a tread about it, and rather handle it there.

Out of the few who watched a demo most were unsure if he was legit or not, and couldn't draw a conclusion.
I had since confronted him about the accusation in good faith (which i now do regret) and this whole problem has just been lying dormant without anything being done.

Being encouraged by an admin i decided to do the last thing in my power and start a tread, and i will link more demos if somebody actually bothers watching all 13 of them.


===================
Moments of interest
This list (and for each individual demo) might be expanded in the future
===================
Note: demo 1-5 are from the same game, but i had stability issues so i couldn't afford to corrupt one should the game crash.
In demo #1
3:26   - minor saw tracking
3:30   - snapping and short tracking (watch in slowest speed to see how he synchronizes perfectly with its movement)
4:35   - pretty apparent wallhack. Use r_showtris and see how he focus on exactly that area where the dretch is and even jumps to reach it.
4:52   - more blatant wallhack. He was not supposed to see that egg.
5:37   - snap and tracking
5:54-8 - snapping

In demo #2
6:41   - saw tracking
6:57   - saw snapping tracking
7:29   - snapping
10:18  - minor tracking

In demo #3
12:00  - possible wallhack on granger and perfect aim while kamikazing down. afterwards he was surprised it was a granger even though it made sound, suggesting he only vaguely targeted it through the wall while the bot did the rest.
13:37  - looks like the bot tried to hit the dretch through the box
14:22  - rant tracking (watch in slowest time to see it clearly)
14:45  - rant tracking
15:00  - goon tracking
15:03  - hardcore mara tracking

In demo #4
15:37  - very nice snap and tracking with luci cannon :D
15:59  - minor dretch tracking
16:24  - hardcore tracking of mara
16:48  - the gold of the game. serious tracking of mara, and a goon disturbed the tracking when it got over the cursor

In demo #5
02:41   - he locks onto the head from a distance and continues till the md dies
- sorta messed up this demo with two map loads :|

In demo #6
04:36-9 - fast pounce tracking

In demo #7
01:39   - snapping and tracking with mara
01:44   - same, and possibly a miss-target at 1:47 (might just be the jerking you see in low timescale between different frames)
02:14   - snapping and tracking with goon
02:40   - snap/track
03:14   - very clear tracking and a much better demonstration of miss-targeting. as he snaps to the middle human the cursor stays closest to him for the entire time while the target moves away, he didn't manage to toggle the bot off and look at the nearest human fast enough. for people unsure about their ability to spot bots this is a pretty easy mistake to see (in slow time!)
04:37   - tracking with dretch
05:35   - minor snap/track

In demo #8
07:53   - minor snapping, a little shaky

In demo #9
09:17   - snap/track on rant and friends
09:53   - dretch snapping
10:03   - dretch snapping, and if you watch in slowest speed you can see the two-speed mode this particular bot uses. once a target gets into view you turn with a constant speed of 10 degrees per snapshot, and when under 10 degrees away it moves only one degree per snapshot

Extra
In demo #1
06:37   - snapping
08:21   - blaster snap
10:18   - dretch snap
10:25   - blaster snapping and tracking...
11:42   - noticeable snap
11:44   - clear snap without shooting

In demo #2
01:36   - mara tracking
02:50   - dretch snap, compare slow time to normal time to see how odd that looks
03:46   - dretch snap
04:52   - goon snap

In demo #3
02:30   - 3 snaps
02:49   - snap
- along the end i got a bit tired, some interesting scenes but not good enough to list it

I will try to add more later but these alone should be enough for someone who knows what a bot (and the new bot) does.


===================
Demo list
This list might be expanded in the future
===================
#1 http://www.mediafire.com/download.php?262gqdj48qqxia6
#2 http://www.mediafire.com/download.php?87uxt7k75ho9jht
#3 http://www.mediafire.com/download.php?op337ng4224xa3n
#4 http://www.mediafire.com/download.php?ycn40ec9h5utyd0
#5 http://www.mediafire.com/download.php?21ne8qjhb8zd4yg
#6 http://www.mediafire.com/download.php?qu547102uy081xq
#7 http://www.mediafire.com/download.php?cg1l35ajcorvr8c
#8 http://www.mediafire.com/download.php?149541op6u4iqd1
#9 http://www.mediafire.com/download.php?276mh8pfp9phehf
#12http://www.mediafire.com/download.php?h5qxa2a244bh4pj

Extra - turned from specking dracone to specking ppnl :)
#1 http://www.mediafire.com/download.php?36i4yjn3m7vb0ym
#2 http://www.mediafire.com/download.php?s4lo7qskz97qo25
#3 http://www.mediafire.com/download.php?dq5qq2ljd28l0on

I openly accept all questions and criticism, but please keep it related to the topic.
nab

Edit 16/9: Followed up on the 8 demos left that didn't get commented. #5, #6 and #8 aren't that interesting but #7 is a good example of alien botting.
Also tried to make the text more readable.
Edit 24/9: Put up demo #12 for proof of identity, will go through it later.

[Nux]

I should start by pointing out that:

• Any sufficiently advanced player is indistinguishable from a bot.
• Any sufficiently advanced bot is indistinguishable from your average player.
• There will be differences between what we see of PPNL's actions through that demo of you spectating him and his actual actions which may make him look more/less suspicious.
• You should always suspect yourself of confirmation bias.

At this point I've only watched demo #4 and I have to say the last few seconds do look pretty botty (where he's flying while shooting the mara with las). It's the way the aim accelerates in strict time with the motions of the target as well as his lack of attention to correcting his position to better view his target as if he didn't actually know where he'd be aiming, more than the actual position of the crosshair that makes me suspect it.

When a human aims, he follows little good habits- based on snapshots of the situation -that have tended to work well. When these imperfect aimbots aim though, they still base their movements strictly on most recent information with zero regard to how you are moving and it looks like this can be seen here.

[CreatureofHell]

I should start by pointing out that:

• Any sufficiently advanced player is indistinguishable from a bot.

Tell that to the obvious feats pulled off by bots that no human can do.

[naboo]

Thanks for replying so fast!
Ok Nux on 1 i will disagree because even with the newest bots there are still traits you can spot in slowmo which cannot be spotted on pro or regular players.
On 2 i will have to agree, with time they might become very hard to distinguish, yet it will take a long time before this becomes a reality.
On 3 i'm not so sure if those differences actually affect the pattern i look for in a botter.
On 4 i try to do that all the time actually. And i also watch players i don't suspect to check and compare if i really can see a difference.

I have re-watched these scenes many times and in different timescales. Sometimes watching in regular time might tell you the full picture, other times you need the full detail with 0.1 time, and also in some cases 0.5 time can spot what the two extremes cannot :D

[vcxzet]

Another pointless discussion
Tremulous has no central auth so no one but the server admins can do

[naboo]

Please keep on topic, i want this tread to be about this issue and nothing else.
And fyi making a tread like this will help inform the active admins about my suspicion (something i might have failed to do earlier).
Also you can get input from players which is very valuable.

[Nux]

There's nothing contentious about my bulleted points: 1. and 2. are truisms; 3. merely states that a difference exists with no mention as to how significant the difference is; 4. is just advice.

I would be weary of using slower timescales. It increases the amount of interpolation frames displayed and can make things look botty that never were. Just stick to normal speed and you'll be fine: you might think slowing things is a good way to keep from missing something but your brain is well adapted to picking up on irregularities in realtime. Slowing things puts you in unusual territory and so can make it harder to notice truly unusual events as well as making the whole process more drawn out which can lead to boredom and poor analysis.

Just a minor point: Why do you keep saying 'tread' instead of 'thread'?

[naboo]

....because i can't spell? :(

As for timescales i don't agree. Yes basically you view it in real time to notice if something seems odd. But certain irregularities can only really be self-confirmed (for my part anyway) by examining it in slow motion, and in a few occasions the slowest timescale.

For instance, something which you just perceived to look "odd" in a demo might reveal itself when you look back at it (in slow time) to be an extremely perfectly controlled dead-on snap and following shot-off but which happened in a matter of milliseconds. In this case timescale would be the difference between being uncertain and being 90% sure. Using timescale to review demos of people who aren't botting will not normally show these patterns.

Of course there are always lucky shots and weird glitches, which is why i base my accusation on a wealth of evidence, and not just a single occurrence.
If you bother, do check out those wallhacks i noted, and there might more clear scenes for all i know...
nab

Edit: and already i'm sounding like an asshole :/

[/dev/humancontroller]

For instance, something which you just perceived to look "odd" in a demo might reveal itself when you look back at it (in slow time) to be an extremely perfectly controlled dead-on snap and following shot-off but which happened in a matter of milliseconds. In this case timescale would be the difference between being uncertain and being 90% sure. Using timescale to review demos of people who aren't botting will not .

WRONG. the client interpolates view directions between the ones in each snapshot received from the server. if the server runs at sv_fps = 20, which is very likely, then these snapshots will arrive roughly every 50 milliseconds. there is no turning behavior with breaking points every 50 milliseconds that would go unnoticed in one of the said timescale settings, but be outstanding in another setting. except if your computer is fucking slow.

OTOH,

can make things look botty that never were.

that's also doubtful.

[Nux]

Edit: and already i'm sounding like an asshole :/

You didn't sound like an asshole to me but even if you did you'd be in good company on these forums.

can make things look botty that never were.

that's also doubtful.

Interpolation between a frame where you're aiming away from a target and a frame aiming directly at it, over an extended amount of time, looks exactly like an aimbot snapping to said target. All that's necessary for this to happen is that you're watching the demo in a low timescale and one of the key frames was over the target.

[/dev/humancontroller]

Interpolation between a frame where you're aiming away from a target and a frame aiming directly at it, over an extended amount of time, looks exactly like an aimbot snapping to said target. All that's necessary for this to happen is that you're watching the demo in a low timescale and one of the key frames was over the target.

no, low timescale is not required. they key frames still get interpolated at unit timescales.
"extended amount of time" doubtfully makes any difference, if coupled with the fact that low timescales are generally used over longer periods of times, during which the viewer should notice the "per-keyframe interpolation" that happens all the time.

[Nux]

Are you saying that the number of displayed interpolation frames between any two key frames is constant regardless of timescale? I don't believe that. Slow timescales are just too slow and too smooth for that to be true. The demo only has so many frames of information stored in it and turning down the timescale is bound to increase the time spent viewing interpolated frames, and hence a higher number of interpolated frames to keep the view smooth.

Seems more likely you're misunderstanding something.

[/dev/humancontroller]

Are you saying that the number of displayed interpolation frames between any two key frames is constant regardless of timescale?

no.

The demo only has so many frames of information stored in it and turning down the timescale is bound to increase the time spent viewing interpolated frames, and hence a higher number of interpolated frames to keep the view smooth.

correct.

you're misunderstanding something.

WRONG.

[Nux]

Then I'm not sure why you disagreed with me here (other than wanting to use your catchphrase):

no, low timescale is not required. they key frames still get interpolated at unit timescales.
"extended amount of time" doubtfully makes any difference, if coupled with the fact that low timescales are generally used over longer periods of times, during which the viewer should notice the "per-keyframe interpolation" that happens all the time.

I know that interpolation happens even in real-time but my point was that when viewing demos in lower timescales you'll be looking at a lot of interpolated movement and you might falsely interpret this as user controlled and hence: "I turned down the timescale and noticed how the aim moved in a perfect line and stopped dead on the target. AIMBOT!".

I'm pretty sure we're on the same page.

[your face]

hey naboo! how's things?

[naboo]

Hi face! Well i guess things are mostly fine :)
Sorta kicking a regular on gpp in the balls atm but i think he deserves it :(
In the end we will hopefully see if it was the right course of action or not (just to be clear, nobody has been banned or anything).

[your face]

i see, lol.  i haven't checked any of the demos out since i am so lazy, but i trust your judgment in the matter.
hopefully see you on aa sometime soon.

[/dev/humancontroller]

in lower timescales you'll be looking at a lot of interpolated movement

correct.

"I turned down the timescale and noticed how the aim moved in a perfect line and stopped dead on the target. AIMBOT!".

"but before and after the aim moved to and stopped dead on the target, the aim moved in straight sections around the target. ANTI-AIMBOT !".

[Nux]

Agreed. The effect is quite minor and I should probably assume more intelligence on the part of the observer. It's mostly just a preference of mine not to fiddle with timescales and if it works for others and they well understand what they're actually looking at then that's fine by me.

I will get round to looking at some of the other demos you've made the effort to produce, naboo. Hopefully I should have time tomorrow and I'll give my perspective on them. For the moment, though, I hope it's clear that I am suspicious and can see why you're accusing him.

[Undeference]

On topic: I watched the demos (except I think I skipped 7) and aside from a few interesting things, I saw nothing suspicious. If PPNL used an aimbot, it was disabled most of the time. If PPNL used wallhack, it didn't help that much. One game someone commented that PPNL's stats went way up on tremstats (really?) and a possible explanation is something I witnessed in the demos: a clear preference to keep the game going to get more kills (e.g., standing in the alien base with a mass driver in karith, trying to kill dretches as they spawned rather than ending the game).

To be honest, I was much more interested in the play of who you suspect was dracone. If it was, he got much better since I last saw him play (hax!), but that was a while ago.

Not: interpolation is constant, so timescale <= 1 really doesn't make a difference unless you are looking at every little twitch (aimbot!) that you wouldn't see at real speed (too smooth: aimbot!). I.e., when you've already got confirmation bias, it doesn't matter how you look at something; it will always support your conclusions.

[naboo]

I just really want to comment on this:

If PPNL used an aimbot, it was disabled most of the time. If PPNL used wallhack, it didn't help that much.

The only difference between using an aimbot or wallhack a little or a lot in a demo, is that it makes it more or less hard to establish whether it is actually used or not. The offense is still the same. Also whether it is useful or not for the cheater makes no sense to me, if it can be proved to be real that's all it should take. And as you probably know, toggling a bot (aim/trigger) is all about using it as little humanly possible to reduce the chance of being caught.

Thank you for bothering to go through the demos.

[Superpie]

I checked out a few of the demos. There is not a wallhack. PPNL jerks off or picks his nose or something while he plays sometimes. If he could see through walls he would at least try to adjust the screen's peripheral vision to scan rooms ahead. There is some tracking but it is not on par with an aimbot as far as snapping, hitbox centering, and wavering from target.

[ULTRA Random ViruS]

The nubmer of 'demoframes' a second is dependant upon what the recorder set /snaps to. It is by default 20. When spectating someone else, the highest you can get is related to sv_fps though.

I'm sure, ppnl does not hack. He compeltely sucks when he comes to asia server where he has 250 ping. I don't think he's ever hit me 100% with rifle before, with an exception of minor spawn camping and type kills.

[RAKninja-Decepticon]

I'm sure, ppnl does not hack. He compeltely sucks when he comes to asia server where he has 250 ping. I don't think he's ever hit me 100% with rifle before, with an exception of minor spawn camping and type kills.

he completely sucks on us1, too.  unless he's improved by magnitudes during my hiatus.

[naboo]

I'm updating the first post with added comments for the demos i couldn't last time.

Also:

There is some tracking but it is not on par with an aimbot as far as snapping, hitbox centering, and wavering from target.

You will find all those in the times i referred to in the first post, and in addition from the update.

Also now that i checked, i started to record ppnl about 7 months ago, and informed him about it a week later.
I stopped recording him after that because i believed there wasn't any more point to it, but later i discovered he hadn't turned it off at all so i kept going. However it got harder and harder to spot over time, along with his score going lower than before.

[ULTRA Random ViruS]

I will post this on the behalf of (no permission) PPNL.
http://www.youtube.com/watch?v=xvl0CvKocGw&feature=youtu.be
He told me that 'frag videos' are meant to show your best moments when i told him about editing video clips. [I don't edit out my deaths or even feeds, but i do fastforward past building times]
Then i told him 'You sure that's your best?'

[naboo]

That's nice (and i think i have seen them before), but the frame-rate is far too low and there is no slow-play function, so its not really helpful if you wanted to show an alternative to my obviously biased viewpoint...

Also i don't really like all the effects and edits in this kind of situation, and its not like there is such a thing as countering demos which proves someone bots with demos proving they don't bot. At the bare minimum you will show that the player has skills in addition to his neat binds.

If you want to post more frag videos you could do it in another tread, unless you really think it is relevant to his case.
nab

[jr2]

Alright, you earned it:

THESE ARE TREADS, why are they in my THREADS?!

[naboo]

FFFFFFFFFFFFUUUUUUUUUUUUUUUUUUUU
i will shove both the treads and the threads up ur hovelhole if i ever see you ingame!

Seriously the tread just snuck in there i don't know how       i blame alien/reptile conspiracy....

[ULTRA Random ViruS]

That's nice (and i think i have seen them before), but the frame-rate is far too low and there is no slow-play function

keepvid.com - download the vid, play it in slow motion.
With the framerate problem, that's his problem. My in-game fps isn't high at all while using fraps. Take this for instance.
http://www.youtube.com/watch?v=BGd1hOH4Af4
I even reduced the video record quality 1/4th size to improve the in-game fps. The fps counter is a joke, because thats what it always say when i'm playing tremor. My real fps is actually half of what it says, i already proved this somewhere else... and fraps fps counter also says so.

If you think ppnl hacks, then this sucky video i just posted is superhax. Seriously, i think a noob like me can beat him easily. Well, i'm not actually 'new', but my aiming skills suck high because trem was in fact the first fps game i could ever play on my other computer. Games like need for speed ran at like 0.5 fps with no sound enabled.

Anyway, let's play the game, not linger on forums. The forums are already being invaded by bots like http://tremulous.net/forum/index.php?topic=16852.0
We need more soldiers. If it even means recruiting you, I'd accept.