I can find the right place to click by finding an image of the box.
Just positioning the mouse and clicking does not work -- nothing happens.
Clicking and holding does not work.
I noticed that when I (personally) move to click the checkbox, the pointer changes to a gloved hand (mickey mouse style) as I get close (~10 pixels) to the checkbox.
I tried KM starting from far away and using a repeated move-1px-relative to the mouseLoc to get to the right spot; the pointer did not change, and the click failed.
I tried KM jiggling the mouse a little toward the end; the pointer did not change, and the click failed.
I tried holding the mouse in the air while KM did its bit and failed, and then I clicked the mouse; I do have a magic mouse, so maybe I moved the pointer a pixel while clicking? In any case, it worked.
Things I haven't tried:
Starting from more than 100 pixels away.
Using pause to slow down the movement of the pointer.
Making the whole path the pointer follows a somewhat random path.
You are entering a cat and mouse game. I'm okay with that. I can't see the code (it's possibly client side javascript) that they use to detect if a person is human or not, but it could be little more than some statistical analysis of the positional data that the mouse generates. Almost anyone who writes code to move a mouse using software will probably not emulate the exact kind of motion that a human hand and arm would have when moving a physical mouse. There are various statistical measurements a web server's code could take to determine if the mouse is being controlled by a human hand. So how do we defeat code that measures human mouse motion? Simple: we simply use KM to record a human moving a mouse from point A to B, then replay that motion later when I want to do something like click on a box. Doing this would likely, but not certainly, fool motion detection software. I've done it, and it worked for my purposes, although my purposes were different from yours. I really enjoyed writing that amazing macro. I'm not sure if I still have the macro because from time to time I just delete all my macros and rewrite everything from scratch. Why? Because every time I rewrite my code it gets better and better.
On a related matter, think about this... the javascript that you get from any web page could measure the curvature of the mouse's path to determine the length of the user's arm. That arm length could imply the age, and maybe even the sex, of the user. (Men have longer arms than women, on average.) I'm not sure if Javascript writers have thought about this or if companies like FaceBook are taking advantage of this. If they can tell a user's height and sex, just from a mouse's motion, how much easier would it be to tell if the user is human or just software?
The macro that I wrote allowed the user (i.e., me) to create and record multiple mouse motions, then it could also be used to slide the mouse over to a new location using a randomly selected path that was previously recorded. It was absolutely hilarious seeing the mouse move like a human would move a mouse, especially when I added some loop-de-loops to some of the recorded paths. I could entertain myself for a long time by putting this macro in a loop sliding the mouse to random locations on the screen.
I'll give you a chance to try writing this macro yourself before I look through my KM Editor to see if I still have it. It was a very rewarding macro to write.
