Developer Popular Post Maxi Posted March 5, 2015 Developer Popular Post Share Posted March 5, 2015 Quoted from: http://osbot.org/forum/topic/66338-reducing-banning-rates-behind-the-scenes-at-osbot-27-febuary-2015/ Mac is supported now On 2/26/2015 at 8:25 PM, Maxi said: Dear community, Prepare for a long but awesome read! I feel like giving you all an insight in what has been happening behind the scenes of OSBot lately and what is to come. We have been working hard in the last couple of months to get the bot back where it used to be in the summer/autumn days of 2013. Fixing issues with the API and the bot is one part of that. The other part exists of creating new innovations and adding new features to keep up with other bots or to make OSBot unique and innovative. These were things we were doing quite nicely back in the days and it's time to reclaim this glory. Many of you must have noticed that MGI was added to the team as a developer. With this promotion we also promised to find a solution for the high ban rates that we, but also other bots, are facing. Ban rates. Detection of bots can be divided in two main strategies. The first strategy is detection on a low level, this means detection of the modified RS client. The second strategy is by analysing behaviour of players. Both can be extremely efficient at what they do. Combined they do a pretty good job, considering the banning rates some people are facing on any bot they use. The first strategy is done by detecting whether the RS client that runs is modified or not and whether it is running in it's natural environment. The way current way most bots work is by downloading the RS client and modifying the client to be able to communicate information of the game with the bot engine. There are ways around it, but the way RS currently has their system set up means that battling their strategies to detect bots on a code level are meant to break and would require a lot of updating. Basically using a bot these days or for example RSBuddy (a runescape helper, not a bot) just plain out tell RS that they are bots. The second strategy is behavioural analysis. By having intelligent systems that compare behaviour of players/bots and compare it to statistics they have of players that are proven to be bots, they spin an entire web of intelligence that could recognise bots. This they can't obviously do for every player, so they most likely have different monitor levels, where each level higher means that the behaviour is analysed in more detail. Those two strategies are then probably combined, where the first strategy is probably the main flagging system to be upgraded to monitor tier 1. Monitor tier 1 then monitors on a global big picture view. Botting times, activities, quests being done, experience gain, trading activity etc. If monitor tier 1 suspects suspicious behaviour, your account will be moved in to monitor tier 2 which might analyse mouse behaviour or other more in depth and resource intensive analysis. From there there might be more tiers until you get banned. This of course is speculation. Having good scripts does lower the chance on detection by a lot. Changing scripts, having different behaviour on the same scripts by two different users, intelligent and human like action etc can all make a difference. But these won't make a difference if a bot basically says it's a bot based on a technical level of the code. So what is the solution? The theoretical solution seems simple. Come up with an idea that has no possible detection on a code based level. The implementation of it however is far from easy. Now to the interesting part: MGI has been working on an idea conceived several months ago and is nearing its completion. We call it the mirror system. The mirror system should tackle strategy one completely. OSBot's new Client Mirror System. Before blabbing about all kinds of technical details here is a picture of what it looks like. A picture is worth a thousand words they say. MGI called this feature the Client Mirror System. It leaves NO trace in the RS client of our bot and is by no means visible on the JVM. The RS client runs in its own environment which makes it even better. You still have all the controls you had in the past. This could potentially knock out the initial flagging system that takes your account in to monitoring. We don't know for sure how their system works though, but this completely diminishes the detection on a code level base. This feature does require additional CPU, so the future will have to show how much of an impact it will have if you use this for big bot farms. We will however provide you with the option to use the old injection system or the new Client Mirror System and they will be able to be used in a mixed fashion if you prefer. In any way, the performance overhead might be worth it for everyone in the end. Sincerely, The OSBot team 28 Link to comment Share on other sites More sharing options...