'Floating' issue with Digispark

I have been trying to use a Digispark to add some smart control features to an automatic toy gun I own, to replace its simple electromechanical contacts. I'm getting unreliable input and since this thing can totally shoot your eye out, I really have to fix this issue. In specific terms I'm getting occasional false positives, causing the gun to fire when I didn't tell it to. The frequency of this issue sometimes increases if I wave my hand close to the device or touch the insulated wire.

Having done some warm-up reading on the subject, I think my problem is referred to as a 'floating pin'. I understand that a pin with no voltage applied to it will not return a reliable false, because of reasons and science. As I understand it, the solution is pull-down resistors which connect my sensing input pin to a GND signal, artificially keeping the voltage low until a higher-amperage high-voltage input overrides it when the sensor button is pressed and HIGH flows through it.

Thinking myself a cunning and creative man, I endeavored to solve my own problem and solder 10k pull-down resistors from my GND pin (1, LOW) to each of my sensing input pins (0,3,4,5), leaving output (2, HIGH) which runs to each of my sensors. To my great consternation, the pull-downs did not improve my situation at all. I was still getting false positives, and after a process of elimination, I found pin 3 to be returning false positives, but no others. Assuming this to be some kind of construction fault, I bought another Digispark (I mean come on, they're $9!). This one is a V4, and has the same issue as my V2 above with the same pull-down setup. My frustration is beginning to take the shape of a very real Blendtec blender. Am I doing something wrong?