I don't want to get into politics on either side, and there are several factors:
- The first is the "doc fix". Basically, there's a 21% automatic cut in the payments to doctors in Medicare. That would lower the payouts to the point where doctors would often lose money seeing a patient, and doctors would stop seeing Medicare patents. Like with the AMT, each year congress approves a temporary fix. It is likely that, while under CBO rules the score assumes there won't be more temporary fixes, there actually will be.
- There are several areas where the bill promises unspecified future cuts. The CBO numbers takes them at their word. Consensus is that these cuts won't actually happen.
- ~$53B comes from the fact that Social Security will take in more money because more companies will pay people wages so they can buy health care on their own instead of the company paying for it. However, it doesn't count the fact that SS will have to pay out correspondingly more.
- The CBO looks at a 10-year horizon. Most of the costs are scheduled not to start immediately, but to ramp way up later.
There are a bunch of other 'hacks' in the CBO scoring to keep the cost near the promised $900bn. Both sides do it, but they discovered a bunch of new tricks this time around. It means that, going forward, CBO numbers on large bills (from both sides!) are probably worse-than-meaningless.
On the older-to-younger q, it's mostly about how community-rating is implemented.