Web3 is centralized and inefficient

Mostly decentralization I think. Instead of going to Cloudflare for CDN, AWS for file storage, everything is hosted on other people's computers for a fee.

Yes: Tor is "somewhat decentralized" in that there are 9 servers run by people to maintain the directory instead of 1, and Roger Dingledine--one of the main people behind Tor--claims to have met in person only like 2/3rds of the participants; but is it as centralized as Etheteum? No. Most "web3" tech (god I hate that term as someone actively in this space :/), BTW, literally are based at least some way--if not extremely literally--on DHTs... including Ethereum, whose nodes self-organize using the Kademlia DHT model. The problem is that you have to figure out some way to deal with sybil and eclipse attacks--which I2P (referenced as somehow good in this article) was so bad at dealing with that it was essentially destroyed by the UCSB security lab, the principal investigator of whom, if you now talk to him years later, stopped bothering to research I2P as they never bothered to really address previous complaints--is that you need to figure out some way to make it hard to join the network (without throwing up your hands in defeat and simply assuming that 9 hobbyists are going to be sufficient to police a network like Tor, which we know is precarious as people keep finding new large, dominating entities to report about every few years), and that's exactly what all the work on cryptocurrencies is about: figuring out the incentive structure for how to avoid these attacks we have been studying since well before Bitcoin pioneered Nakamoto probalistic consensus (I was doing distributed systems research in like 2003 and we had been talking about these attacks for years earlier even then) and then how to protect and come to consensus on that incentive system itself (as you rapidly end up figuring out you need some kind of tradable and storable reputation system... aka money). The reality is that cryptocurrencies are the most recent culmination of distributed systems research, and the extent to which they are inefficient sadly is largely because humans simply have not yet figured out how to do it better (and if you really think you can, please do instead of calling the people actually doing this work somehow stupid for not doing the "obvious" thing of attaching a DHT up to a branch and bound or map-reduce computation system, as that was the work people at UCSB were trying to do back in early 2000s and you know what the professor who was in charge of that lab is now super excited about? cryptocurrencies, as he knows they solve real problems he ran into trying to build a decentralized computation platform). (And to be clear: do I think Ethereum or even Avalanche has solved all the issues yet? Hell no. But they are going in the correct direction... if--and this is where the reasonable arguments about cryptocurrencies tend to be--you believe decentralized systems are actually important to achieve or should ever be used instead of a centralized one: some people seem to disagree with this--which I certainly disagree with, but I don't think is entirely crazy--and that leads them to constantly want to replace cryptocurrencies with centralized systems, but you can't point at older decentralized systems and somehow claim they are better than what we are building today without first analyzing why they never really went anywhere and asking how and where--and then why--cryptocurrencies and their distributed blockchain tech are actually even that different.)

This seems like the angry ramblings of someone who missed the boat.

I'm not invested in cryptocurrency or other tokens, but this seems very unfair to the ecosystems that power them. Sure, there are centralized wallets, marketplaces, APIs, but that's just not the point, not the problem they're trying to solve.

If Google deletes your account, that's it. If OpenSea deletes your account, another app will happily connect to your wallet and let you continue where you left off.

> Look at email, while the protocol is technically “decentralized”, but look at how big Google Workspace and Microsoft 365 is. Sure, you can easily set up an email server on a $5 VPS, I do it with my domain, but most people just leave it to big providers. Web3 is no different.

This is an interesting argument to make but I wouldn't say having large subsets belonging to single entities equates to it being "not decentralized."

When a product has a lot of negative externalities, and you also notice that it doesn't deliver on the promise that's meant to counterbalance those externalities, it's entirely reasonable and appropriate to conclude that the project should be rejected, rather than joining in to try to fix it. I don't know why anyone would want to build a "Web3" with people who use "have fun staying poor" as a catchphrase, or whose understanding of existing technology allows them to refer to "BSD linux". I don't want to empower or enrich those people. We already have enough of them.