Both reliability and performance. And also reliability is something which you estimate, calculate basing on some models - it's probability, not a guarantee. How

Both reliability and performance. And also reliability is something which you estimate, calculate basing on some models - it's probability, not a guarantee. How would you estimate the reliability of an engine which run over 3 hours on a test stand? Of course you might prefer a well known fuels, industry-proven range of pressures and other engineering solutions, and you test them on Earth first - but after that you have to fly. Track records are tricky for rocket systems. Since the first Sputnik, over the 57 years the humanity had may be - roughly - 4000 launches of space rockets of all kinds, with many different fuels, engines, sizes etc. It's relatively rare when you have a good track record which is measured by many tens of flights - let alone hundreds. When a system flew 20 times it's already considered rather mature - double so if those were all successful flights. So it's quite usual to use a new engine in a new design of rocket - in fact it's rare when you have many different rockets with the same engine type or many different engines used on the same rocket. Rockets and engines are usually optimized to each other - particularly rockets are built around engines, since margins are quite thin.