I have attempted to roughly replicate the intended probabilities of the original mechanism, but with much less dice rolling, as I personally think befits a Sword & Sorcery version of 5th Edition D&D. Basically, the idea is that you are
already generating enough randomness, so why not use that instead of asking you to roll even more dice?
You have a 1-in-20 risk of getting a failure, which is 5% much like the original idea, except it actually is a failure (no further steps needed) and not just a Bane-like effect (that carries a significant risk of having no effect = you still succeed).
You have a 1-in-20 except 1-in-20 chance at getting a success, which is 4.75% much like the original idea, except it actually is a real success and not just a Bless-like effect (that carries a significant risk of having no effect = you still fail)
I haven't done any deeper mathematical analysis, but I am prepared to argue this accomplishes much the same effect as the original playtest proposal in practice, yet with a minimum of fuss. Should anyone crunch the actual numbers and find out I am wrong I would welcome you sharing your results. I am well aware that hunches and guesstimates often mix badly with actual probability analysis!
Note how I personally think the mathematical average of spending your bonus action should be greater, but that I haven't let that influence the above proposal.
Cheers
PS. I don't think I need to say this, but just to bring it out in the open: anyone should feel entirely free to reuse or steal this idea without feeling a need to credit me. DS.
PPS. And of course, should Aklatu be associated with different numbers, you should tweak those. For instance "missing by 4" and "rolling a 17" just to pick one example. As long as the first number is in the single-digits and the second in the double-digits I
think the math would stay recognizable.