SRL Archive - News and General

Hello, im a member of RiD and im looking forward to trying this bot/these scripts. And im just wondering how safe it is to use? ban-wise… Are the scripts some of those scripts that clicks the same pixel(color) every time or is there random mousemovements etc?

Tell me about the bot and how it works :)

We actually got antirandoms instead of just logging out ;) That beats RiD, especially at their insane prices.
And any decent script actually randomizes everything, even most of the beginning scripters do, and good scripts like MSI are pretty human-like. It gives you a "unique" botting style (read: everything is random). And you can actually take a look to see how safe it is, security thru obscurity has never ever worked.
Only thing SRL lacks is the flashy flash swf with tons of GP piles and quotes and vids and such.

Yep, we're all about mimicking realistic human behavior in every script. Another big difference is our scripts can run minimized, NOT taking over your mouse and keyboard while running. Above all we're more of a learning community, in no way limited to making scripts for Runescape. You'll find out we're here to help you and get you involved as opposed to RiD's way of selling the amount of time you can use their scripts and hoping they work. Did I mention everything here is open source?

Okay, thanks so much! but how does the SIMBA client handle random events?

It just looks at the screen and solve them? It can't solve everything but it does solve around 50% I'd say.

Does it log out or get stuck if unable to solve it?

Log out of course, we have been around for over 5.5 years, no nooby getting stuck :p

I don't believe SRL's mouse movement functions are anywhere near human-like based on my analysis, but I haven't seen RiD's.

From: Aftermath
I don't believe SRL's mouse movement functions are anywhere near human-like based on my analysis, but I haven't seen RiD's.

Benland has done extensive research on the mouse movements. They are pretty humanlike, except for the fact they don't overshoot from the target.

From: Aftermath
I don't believe SRL's mouse movement functions are anywhere near human-like based on my analysis, but I haven't seen RiD's.

Well, then show me a bot that has/had a better mouse movement (not including RiD).

There was a thread on MITB about mouse movement, I think he's referring to that. However, I still doubt that Jagex tracks all your mouse movement AND send it to the server for processing. Perhaps they do client side processing.

Well, if you think it's not human like, try writing a function that, given a set of mouse movements + time, determines whether it is WindMouse or a human?

I'd imagine click points are more tracked than mouse movement. I remember seeing such packets in earlier Runescape clients and I sincerely doubt they've removed those. ;)

From: Wizzup?
There was a thread on MITB about mouse movement, I think he's referring to that. However, I still doubt that Jagex tracks all your mouse movement AND send it to the server for processing. Perhaps they do client side processing.

Well, if you think it's not human like, try writing a function that, given a set of mouse movements + time, determines whether it is WindMouse or a human?

I've already thought of doing that, and may when I have the time. However, when you consider that a human can tell the difference with no difficulty it isn't necessary to write the program to demonstrate that. You can judge for yourself; here's output from a test program I wrote while working on mouse routines:

This is WindMouse:

This is me moving the mouse:

WindMouse shows a lack of understanding for (or analysis into) how humans actually move a mouse, and the whole concept of a global mouse function to satisfy all mouse movement is flawed in the first place, since humans make different types of movements depending on the task at hand. For example, the movements where you want to get to something far away in a static location (e.g. checking your stats), clicking on something where precision is not needed (e.g. your inventory items, which have an enormous hitbox), and clicking on something where precision is necessary (e.g. selecting an option from a right-click context menu) are very different. [There is also accidental mouse movement and idle mouse movement, which are not properly simulated either].

So, to recap, you need to support different types of mouse movements based on intent, precision, and distance involved.

I haven't looked at the WindMouse source, but honestly from the picture it seems as if it's just randomly picking a distance to skip between points while far away, and then slowing down to a crawl as it's right nearby [and of course, you mentioned the overshooting issue]. The shape of your curve is decent - the distribution of your points along it, however, is very inappropriate.

Well, I thought I'd go ahead and do it programmatically - here it is with a very rudimentary detection method - analyzing the speed function (the distance between two consecutive points).

WindMouse:
Points: 47
Inflection points: 33
Pc: 0.7021276595744681
Count: 41
Mean: 12.133058776409438
Standard deviation: 199.34427385594793
Probably a bot.

Me:
Points: 103
Inflection points: 53
Pc: 0.5145631067961165
Count: 74
Mean: 7.053037783521785
Standard deviation: 1816.844693791844
Did not detect a bot.

WindMouse is unnaturally regular; I could detect it this way on any decently-sized movements. Of course, the acceptable standard deviation threshold varies based on distance (smaller deviation is less bot-like over shorter distances).

Edit: The points / inflection points / pc stuff is from another method which I tried to use first - counting at what percentage of the data points the speed started changing in the opposite direction (e.g. from increasing to decreasing); however, I don't think that would work to differentiate in this type of motion between me and WindMouse. The detection I have relies purely on standard deviation at the moment, and is just a single threshold (which should be modified to scale depending on the distance of the movement).

Edit 2: However, I believe a more reliable detection mechanism would be to analyze the point groupings, since by stuffing everything into standard deviation we are throwing away information. Notice how my points tend to bunch up, for example, while changing direction of the mouse, while starting off, and due to other factors such as the surface the mouse is sliding on, while WindMouse does not bunch at any point other than the end.