People are mentioning Bell Labs, but at it's peak, Bell Labs was THE place to be for condensed matter (my area of physics) research. Academia is littered with people who worked at Bell labs during some part of their career. However, this was not really a risk. Bell Labs offered incredible funding for people and you had REALLLY brilliant people around you. In other words, it was quite competitive with academia and the only question for someone leaving who wanted to go to academia would be whether or not they would be good at getting grants (not a question of their research record). So, if you were a young researcher, starting at Bell wouldn't be a risk, it would be more like getting an offer from say MIT/Stanford without teaching or grant writing responsibilities.
This is a different scenario. YCR is just getting started, so there is definitely a risk involved for a young researcher. In academia, it would be good for a young postdoc to have a portfolio of projects. Maybe one that's likely to result in success, even if it's not groundbreaking and something else which is more high risk, high reward. As an advisor, I would be irresponsible if I didn't try to suggest such a strategy to anyone working for me--that way, even if the risky project fails, they have something they can show that they accomplished when they look for their next position. If YCR offers the researchers the chance to balance working on risky and not so risky projects, then it sounds like a good opportunity for a young researcher--they get funding (it's not clear what the time scale is--a postdoc in physics is 2-3 years) and a chance to focus on their research and to work with outside researchers. Typically (in physics), postdocs don't teach, so that's not an issue. However, the OPs worry that they won't be able to balance their portfolio is a reasonable fear.
Finally, on the issue of publication, this varies from field to field. I have reviewed papers from PRL, Nature, etc. At least when I do it, it's not the same as just writing a comment. One paper that I refereed was a methods paper. This paper will eventually find it's way into a "black box" computer program, so I thought it was important to be correct. So, besides looking in general at the method, I went through every step of the derivations, checked the integrals, looked for sign errors, etc. This is time consuming and a much different process from just writing a comment or two. A good journal does a lot to try to improve the signal/noise ratio of material. Mistakes are still made and sometimes the process breaks down. It could be improved--but
One of the great things about YCR is that it provides another route for people and it's important to experiment with different approaches. Until it's been tried, we simply won't know if this model works or not and I applaud them for putting up the money and resources to try the experiment. Also, the grant process has become rather broken--but that's a problem of there being a lot of excellent proposals out there (I've reviewed grants before), but not enough money out there to fund many of them. Again, I've heard cases where the process has broken down based on politics, but more of what I've seen and heard is that reviewers see a LOT of things that they would like to fund, but there's just not enough money to go around. So, researchers end up spending a lot of time submitting multiple proposals to multiple agencies because the chance of success is too low....
I hope YCR works out!!! But, I do hope that they allow their researchers to do some hedging by working on some incremental projects as well as high risk projects.