We need to put Internet Archive on IPFS or something similar, finding a single location for 50PB+ archive is going to be difficult but one can easily find 50000 people with 1TB of free space and a stable server+internet connection.
That's not how this works. There's currently no truly decentralized storage that could handle the IA workload. IPFS certainly doesn't cut it. People have tried. It didn't work.
Not to mention I'd take a single, stable, commercial Internet link over 50k consumer Internet links where you can't even tell if the other side is temporarily shut down, the network got partitioned/congested or the hard drive got wiped.
[1]: 25% denser than TLC
https://github.com/pirate/internet-archiving-talk
I think they've thought about IPFS and BitTorrent in the past, but 50PB is a lot of data and they already have a few redundant backups around the world that are hard enough to manage as-is. Hopefully the latest round of lawsuits and public support pushes them more in the direction of letting the public help them with hosting.