Now, the 204 year bit sounds impressive, but it isn't like a piece of DNA lasted 204 years without any decay. Instead it was copied repeatedly over that time. If I copied that 4TB hard drive once every 25 years (generation time) onto a brand new drive (assuming that you could keep making them compatible) I don't think that getting the data across 200 years without any bit-flips is really that tall of an order. Sure, technology will change, but that really is a different matter, and I doubt that any commodity computer technology used in the next 200 years will do any worse than what we have today.
Actually, it's more than copying the drive once every 25 years, it's making a copy of data on the drive many times each day -- some where around the 100,000th copy of the drive randomly choose a copy to keep and start the process over again. With that kind of usage on a drive, the failure rate (let alone error rate) will be _much_ higher.