Hmm, reward myself after a fixed interval of 30 minutes? That's just crazy enough to work! (I have heard of the Pomodoro technique before, and I'm not quite sure why I didn't just go for that at the start.)

The hidden random timer is to make myself resilient to extinction and ingrain the habit even without reward. Although, randomly choosing to reward at the end of pomodoros would work too. IIRC, intermittent time interval is the reward structure that survives the longest without extinction, whereas a variable ratio reward structure creates the most vigorous workers.

Also, I think what you describe is a conditional reinforcer and not an intermittent one. What I mean by that is after a long enough time, they subject would become attached to the coin flip itself as a partial reward. Kind of like clicker training for animals, or how a shot at a jackput pull is a reward even when it doesn't payout. Then you could use the stronger conditional training systems...

Your suggestion is brilliant. Aaaand now I've got "write a gamblerdoro app" on my to-do list.