A common hopium promoted by crypto people is “we must decentralize everything and put it on a blockchain” which has resulted in bizarre memes like “web3”. In reality, public blockchains are only meant for very specific usecases, when both of these criteria are fulfilled:
Peer-to-peer operation: software that can be run by anyone, and messages are passed directly between them
Strict global consensus: all peers must agree on exactly the same results
As such, blockchain apps are a niche subset of “web2”, not a new thing, and not something that can or should “replace” “web2”.
Firstly, not everything needs to be decentralized. Most services are best operated in a centralized manner. Further, in a vast majority of cases where decentralization is indeed beneficial, peer-to-peer operation is enough to achieve decentralization. In others, local consensus is all you need; global consensus is not only unnecessary, but also wasteful.
A case where decentralization was greatly beneficial, and it was achieved without requiring blockchains: Alliance for Open Media. Backstory: a private company, MPEG-LA, requires licensing for all media content on the internet - music, videos etc. While open source alternatives did exist, they were never adopted due to the lack of infrastructure. So, a large consortium with many companies and independent researchers collaborated to create an open standard for all. The result is AV1 and AV1F - an awesome public good. There was no need for global consensus: local consensus was adequate for this consortium, basically a real-world decentralized organization.
Till date, there are only two applications where both criteria are necessary: money & identity. Now, it can be argued that there are many applications of money and identity which only need local consensus, but that is already achieved by traditional systems - though they can be enhanced by technology like smart contracts. The appeal to public blockchains then becomes when there’s truly global consensus and access.
Of course, money can take different forms - non-sovereign money, representations of real world assets, governance/ownership tokens, collectibles, loyalty rewards, social rewards etc. As can identity.
There are certainly interesting applications possible that combine both money and identity.
In most cases, though, it’s wasteful to put most things on a blockchain. But there’s also the possibility of hybrid solutions, something I called “cedec” (centralized/decentralized) in a post a while ago.
Let’s consider the case of a “metaverse world”, or basically just an open-source massive multiplayer game: you have a 3D reality where people participate. It’s worth noting that this is not a traditional game made by a company like Blizzard or Bungie. As far as games are developed under the traditional model, you only need local consensus. So, a consortium of game studios, engines and service providers can build standards and operate them. Portability outside of this environment will be largely useless anyway (with certain specific things made global - more about this later).
So, we’re looking at a new type of game that’s developed open-source with many forks etc. Here, the 3D stuff can be rendered, zero-knowledge proofs generated, physics engines computations etc. locally or via cloud streaming, and most/all presence, data and physics can be done peer-to-peer; specific types of data using solutions like IPFS etc. Anti-cheat software is all you need to guarantee local consensus, no need to go through a blockchain.
However, each user can have a global identity, you may opt to give governance rights to the players, revenue sharing with developers, and there can be some element of the game - perhaps a cooperative time-bound physics or narrative puzzle - that requires strict global consensus.
Let’s take the hybrid model one step further: you have a centralized company that wants to run a fork of the above decentralized game, but nevertheless participate in the broader ecosystem.
So, this game is mostly done locally, on cloud, or peer-to-peer; there are some elements that need blockchain-type solutions. The company can run an application-specific volition, where all of the data is centrally held, but specific higher-value transactions and assets can optionally post data in a committee, or even on the L1 data layer for full rollup guarantees - e.g. collectibles in an ERC-721 standard. Crucially, even if all the data is centrally held, the worst-case scenario is assets frozen, not seizure or theft - if the company is denying you service this is a moot point anyway.
In this manner, you can have a hybrid solution that truly maximizes the best of every component and mini usecase in the most efficient manner possible.
Are there totally new experiences possible and make sense outside of the scope of money and identity? Sure, but thus far, I haven’t seen any, it’s wasteful using public blockchains for anything other than those. As mentioned before, what I can see is innovative combinations of the two, and these enabling new usecases even if the product’s mostly otherwise centralized, or decentralized but through non-blockchain methods.