Slitherine

I like the idea of a max R&D reliability being less than the max possible reliability, but I don't entirely like how this system works.

Example:

Maximum R&D of the gemini landing capsule is 76%, its overall reliability is 82%.
Through experience and launches I raised my gemini reliability to 79%, but then had a mission failure that brought it down to 64%. Currently, through R&D I can raise this reliability to again just 76% maximum, which doesn't make sense.

I'd imagine it realistically being a case of finding out what failed and fixing it, thus being able to bring it back to those 79% it had before. Being able to bring it to just those 76% again doesn't feel realistic, its the same design so why should the "experience" be lost? Indeed if anything, some small experience should be gained even through failures.

What I mean is that the "experience-based" reliability, should be a constant and not disappear after a failure. Mission failures should affect only the R&D reliability, not the experience-based one.

I would even add a small bonus on top of that for the same reason. Every failure and the following fix/improvement should result in a higher base reliability (even without the experience from missions, since the failure - even on the first try - would result in some seasons of fixing and therefore in experience and improvements). I think this is also very important for the gameplay. Failing 20 times and fixing for 2 years every time and always getting back to the same base chance is just frustrating while improving the design after every failure would be rewarding and give the player (and the research team) a well earned feeling of progress and achievement. It does not even have to be much. Maybe a persistent improvement of the base max R&D value of around .5 or something. That way you know that every failure will result in progress which gives you motivation to try again and again because you know every time you fix something and your chances improve, instead of knowing... ah well I'll fix everything AGAIN and then be at the mercy of the same odds, no matter what I do, so at some point I just quit.
The chance of the mission being successfull would only be marginally increased by a .5 improvement of the component for every failure... but knowing that every time you fixed it moved you that much closer to the eventually successfull mission is a lot more rewarding than knowing you went through the motions and researched it up to the same odds 20 times and finally beat the always same numbers. It makes all the difference between the player achieving something through failure and the game being frustrating I think.

Yeah, that's something that was implemented into Race Into Space (http://www.raceintospace.org/), the "modern" open source port of BARIS, and I think it is a good thing.

Also, let's face it, the reliability/safety gain for live test flights is too low. Raising it to the maximum would require an insane number of launches, and there is no hope to reach it given that you will always get a failure before that point.

Now, of course, some balancing of the astronaut/mission controllers skills could change things too. Currently, astronauts are mostly a "negative" factor, which does not render justice to them.