The "production" lift capacity included some assumptions apparently about how much they could get out of Raptor and what they expected the assembly to weigh. Engineering constraints requiring more structure, the heat shield being inadequate, and the inability to raise the chamber pressure on Raptor to get the promised ISP have all impacted what the "expected" lift to LEO/GEO will actually be. Don't misunderstand, I am impressed as heck with SpaceX's engineering team and they are definitely getting closer to the point where they will have the design space fully mapped out and can make better estimates. The NASA documents are a better source of news on how Starship is going (as it's slated to be part of the Artemis program) than SpaceX marketing (one is engineering based, one is sales based). AND New Glenn isn't "fully" re-usable, its another 'upper stage gets consumed' platform (like Falcon). That is definitely an advantage with Starship if they make that work. For history, the shuttle has a similar history of shooting high and then finding that the engineering doesn't work.
The first version was supposed to launch 150 tons to LEO. In reality it was something like 15 tons. Even the new V3 (significantly taller) only aims for 100 tons, and whether they achieve it is still an open question.
What do you mean here? I was under the impression that it was increasing each new version. Is that incorrect?