The first is a retroactive public goods funding system, sponsored and partially designed by Vitalik: https://medium.com/ethereum-optimism/retroactive-public-good... . The idea is that, instead of trying to figure out which public goods will be worthwhile in advance and funding those, they collect money from transaction fees (or something) and pay it to public goods projects that in retrospect were beneficial to society. ("The core principle behind the concept of retroactive public goods funding is simple: it’s easier to agree on what was useful than what will be useful.") In theory, you could have a startup whose business plan is "raise money from investors, build a public good, then get a ton of money once the public goods funding oracle agrees our good is beneficial and repay our investors". That's not a change that requires a blockchain or a cryptocurrency or whatever, but it definitely meets the criteria of "some people interested in and working on blockchain and related technologies are pursuing ambitious goals that could reshape the economy and politics".
Next up, Ethereum Name Service. The idea here is basically just DNS as an Ethereum smart contract. This, combined with the really ergonomic key-management tools that have popped up like Metamask, could enable an interesting future: one where everyone has a keypair that's associated with a human-meaningful name. You could imagine websites where your account is a public key, and you "log in" by solving challenge (decrypting something encrypted with your pubkey), and your "username" is just whatever ENS name is associated with that pubkey. Basically, it decentralizes the concept of an authentication/user accounts. (disclaimer, I got some ENS token for free when they did an airdrop that I haven't bothered to sell yet so I guess I have a stake in it)
Third, filecoin/IPFS/IPNS. IPFS is a protocol for retrieval of a file corresponding to a hash (basically it's bittorrent, nothing really that new or exciting, although it performs quite a bit better). IPNS is a way of creating an identifier that's associated with a hash you can update, basically replicating the concept of static websites that the website maintainer can update, except it's decentralized so anyone can provide hosting for the site instead of just the original site owner. (Imagine if you could download every site in your bookmarks and seed it, so they would never go down for anyone in the world as long as your computer is online.) You can associate an IPNS identifier with an ENS name, which is the only way I know of to have a website with a human-meaningful name where the whole stack is fully decentralized. I think IPFS and IPNS feel delightfully "webby", there's no complicated blockchain or anything like that, it's essentially just a souped-up bittorrent.
Filecoin is interesting because its goal is to act as an incentive layer for IPFS. On the traditional web, hosting a website costs money that the website owner has to pay, and then website users enjoy the site for free. That's nice, but it means that without ads or some other form of monetization the website can be a financial burden on the owner. (it's fine for text sites, but imagine a video-hosting site or something, the bandwidth alone can be extremely expensive.) Filecoin rebalances the incentives by allowing the users to pay a small fee to the host (and it's decentralized, so anyone can be the host if they choose), that way as long as there's some interest in the site someone is at least being compensated a little for storage and hosting. I don't know how much progress they're actually making towards any of these goals though, it doesn't seem usable in its current state in the way IPFS or ENS are.