I thought you were questioning on whether or not they were only using MouseKeys or some bot, and them getting unbanned helps strengthen their argument against them using bots --- Double Post Merged, Feb 3, 2016, Original Post Date: Feb 3, 2016 --- I do believe that efficiency is definitely a factor, especially long periods of time with no efficiency drop or change.
I find it very hard to believe because efficiency can't possibly be a reliable heuristic, because not only bots are efficient. D:
I believe the point is that humans can be efficient, however to maintain that efficiency for extensive periods of time is generally beyond us. I recall someone who made an Agility bot for another client a year or two ago who had spent the majority of his time developing a system that randomly degrades the performance of the bot over time, I can't say for certain but I don't believe that bot was subject to many bans if any.
I personally believe it's a tier based system with ever escalating analytical designs. For instance, after doing something for so long you're escalated to the next method of detection, this would require much less resources as the intensive detection could be reserved for those above a certain threshold. If they're storing data for every characters skills and experience it's not unreasonable to believe they could cross reference that data to flag you for more advanced systems. I believe you're under the impression that they would always be running all detection methods on everyone at the same time, but it would me much more likely to run them in real-time with the 50k or so players they have logged in at a given time and analyze them further if they find something mildly suspicious.
There is a simple solution to that. They only need to monitor about 100 accounts at a time. When you get reported they put you in the queue, then monitor you for an hour or so, then move on to the next ~100.
Exactly, it's hard to take this guy seriously if he doesn't even understand the concept of a queue. Not to mention they could run the base level detections on each worlds server for each area so it splits the workload evenly.
Splitting the work load evenly isnt even a thing... the whole "system" is automated... Update 3, still going strong... 16 hours, 90,000 energy and 1 level gained (At this point my account has been logged in and actively botting for 8 days WITHOUT a break)
Just because it's automated means the workload can't be split? So they can't run simple detection on each world and escalate it to a bigger system for further inspection if they deem it suspicious? Why can't this be automated? How do they automate appeals and the bot busting twitch streams? One guy said he's into robotics, I'm very interested in how they make artificial JMods to stream those, maybe he can clear that up for us. Evenly might not have been the best choice of words but what I meant is they could have a simple detection for each world and only store and analyze the data past a certain threshold.
In order for this so called Threshold to exist... all of the data must of been analysed to begin with, otherwise how would this automated system know that the data is supposed to go into the threshold... You are saying that they are using a set of parameters, without first storing the information in which to allocate the parameters too..
All the data they need to store is already stored. Hover over a skill and look at the experience per hour, or the total experience. That's not coming from nowhere, I'm not saying they're not analyzing to begin with, but there is a difference between analyzing something simple and escalating it to analyze more and analyzing everything all at once. Processing something that flags everyone over a certain exp per hour can be done with minimal power and can be done on a world to world basis, which can then be escalated onto a server dedicated to further analyzing, once this is done they can flag that persons account and begin logging more. They don't have to log everything from everyone all at the same time, because as you said it would require a lot. It's the difference between a sample size of 50k and a sample size of a fraction of those 50k who are earning over a set amount of exp. It might not be experience but the point is you start with something simple and go into more complex stages everytime they meet a criteria.
I disagree with this "theory", as it is so easy to tell if someone is macroing, and would be just stupid from Jagex part to not use this as in advantage. No, not all accounts are under the botwatch all the time, but when they are their actions in-game are being monitored. I'll just leave this here, something that I used on my RSPS, and it did work better than I thought. "Fitts's law (often cited as Fitts' law) is a descriptive model of human movement primarily used in human–computer interaction and ergonomics. This scientific law predicts that the time required to rapidly move to a target area is a function of the ratio between the distance to the target and the width of the target.[1] Fitts's law is used to model the act of pointing, either by physically touching an object with a hand or finger, or virtually, by pointing to an object on a computer monitor using a pointing device." Source Wikipedia under Fitt's law Fitts's law - Wikipedia, the free encyclopedia
Well unfortunately my session timed out at about 17 hours of botting so I had to turn the bot off and back on again but its now been running again for another 4 hours and this time i've set it to gain exp instead of just convert to energy. I will leave it running over night again and post another screenshot if its still active in the morning
It can be a reliable heuristic when used in tandem with other heuristics. It definitely wouldn't be the only heuristic they use. I assume they'd use a vast amount of them
Well since people got banned for that specific heuristic (probably more, but I'm guessing this was the trigger), it seems like it's a rather large influence in the pool of heuristics.
So as I said earlier, unfortunately my session timed out at 17 hours so I had to relog, so here is update number 4 at 19 hours and counting. This puts me at the same location.. on the same world for 36 hours straight now and I am still going.
I'll stick with Arbiters theory: "The tl;dr to bot detection is that Jagex has developed (or can develop at any time) models for how humans behave (acceptable ranges for online time, algorithmic estimates for effects of tiredness, behavioral similarity to common botting practices, etc.) and have each of these models act as flags. Set off too many of these flags and you get the banhammer. P.S. Oversimplification for verbosity's sake. Surely the flags are weighted, ranges statistically distributed, etc. As a disclaimer, these explanations are purely speculative, but if I know Jacmob half as well as I think I do and I am half as intelligent as he is, this is the way he would do it because it's the way I would do it. P.S.S. All this being said, I would imagine most bot clients aren't even getting into the behavioral analysis portion. There is probably a tier one non-intensive analysis that is done to catch the clients that have fallen for Jagex's traps and/or utilize hacks/shortcuts that are easily detected. That would explain why certain bot clients are being called out by name by Jagex and accounts being banned simply for logging in through the client without even running a bot."
Just a small update as far as this goes, I continued to bot Divination until Feb 10th (none stop, without breaks) and have made well over 100m from selling the energies I've botted, all from exactly the same location. I have also been botting RC since the 10th and have currently gained over 3m exp in the skill from botting runespan and am STILL not banned.
