Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

ChatGPT gained more daily active users than any web3 company ever will, and the year isn't even over yet.

This is THE crowning example of why web3 has no usecase even though they have been 'building' and it is 'early days' for almost 15 years.

I don't know anyone using filecoin, but I can bet $100K the person next to me knows or has used ChatGPT.



While I don't disagree with your stance on cryptocurrency companies, comparing them with ChatGPT, the latest sensation, is a clear example of survivorship bias. There will always be a super successful company at the top of the market.


Bitcoin was allegedly all the rage in 2010s and almost nobody uses it for payments at all and at scale.

It failed.


A solid half the random convenience stores and gas stations I visit have a bitcoin vending machine. Of everything in crypto, it has failed the least.


Have you ever in your life seen someone use them, though? That seems like this generation’s CueCat but I at least used mine to inventory my books and CDs.


Thanks for letting me know I’ll stop using it now


Active users doing what?

It’s trendy sure, but sparkling autocomplete has a limited number of use cases, and zero shipped products.


Copilot is a shipped GPT product with paying customers.

Who go so far as to never shut up about it.

And make me sit and watch them fumble with it for minutes in screen sharing codepair sessions as they construct lengthy prose descriptions of trivial code snippets to coerce it into being "helpful".


GitHub copilot has (likely) more than 400k subscribers. $4m/month isn't too bad.


Given that ChatGPT has one of the fastest growing user bases of any piece of tech ever (there’s been an infographic floating round of it compared to iPhone, Spotify, etc), it’s hardly a fair yardstick to measure popularity against


It is a fair comparison.

Even if we go back to other AI projects: (DALL-E, Stable Diffusion, GPT-3, LensaAI, etc) all of these projects have been used by hundreds of millions of real people, in a short space of time.

I have never seen a crypto project with a clear usecase ever go mainstream and we are still waiting for it.


100s of millions is a stretch for sure.


15 years is early days when the technology is completely novel. For reference: the internet started in 1969. It took 30 years for it to see mass adoption.

Ethereum is extremely primitive at the moment. You reveal your address, and all of your transactions, to the entire world, when you use it. Scalability of L1 is limited to 30 transactions per second, while L2 is still in beta, with centralized coordinators.

With more development, people can conduct transactions while maintaining total privacy, and the public blockchain can, with the use of temporary data layers (EIP-4444 in Ethereum), sharding and Rollups, process 100,000 transactions per second while maintaining the decentralization, tamper-resistance and permissionless-ness characteristic of blockchains.


The difference is that the internet was immediately useful to people whose jobs weren’t selling it, and adoption was held back by severe constraints (computers and networks were slow, limited, and very expensive) until the mid-90s. Despite that, the web started changing the world within a couple years of the first release.

Ethereum’s had more time where it’s had global availability at much lower prices but it’s hard far less impact because it doesn’t solve problems almost any people have.


Ethereum is a financial platform, so its uses are going to be different than the internet's. So far, it has been used by the Ukrainian government to do fundraising:

https://futurism.com/the-byte/ukraine-selling-nft-military

And to buy weapons/medical-supplies:

https://twitter.com/nicksvyaznoy/status/1605248783192887298

It's used by artists to make a living:

https://www.cnbc.com/2022/02/02/this-36-year-old-made-over-1...

And it's used to transact with over a hundred billion dollars worth of stablecoins, in countries like Argentina where the currency is unreliable:

https://cryptonews.com/news/stablecoins-lead-charge-as-crypt...

And this is when it's in a primitive state, with very limited scalability, and zero privacy. Once the technology is more developed, to allow something substantially greater than 30 transactions to be processed per second, and to allow people to conduct transactions on the blockchain without revealing their financial assets and activity to the world, it will likely gain more adoption, especially in the realm of stablecoins.


Yes, I’m aware there’s been some niche usage but it’s still quite limited because it doesn’t outperform the alternatives on almost any task - yes, it’s cool that Ukraine gets so money but it’s a tiny fraction of the total. The real question is how much margin you can save: anyone can transfer money, the question is whether you’re cheaper, safer, etc. If the pitch is “like PayPal but 10% cheaper” that really caps the total growth potential.

Similarly, the privacy things are both a huge change and something only needed due to having built the system on the wrong architecture. Solving that problem means very expensively reaching the point where its competitors started.


>>Solving that problem means very expensively reaching the point where its competitors started.

Adding privacy makes transactions on the order of 4X more computationally expensive.

Once Ethereum has EIP-4844 and EIP-4444, such a transaction would cost less than a penny in fees on a Rollup. Stablecoin payments would be substantially better than centralized e-wallet payments for many mainstream use-cases if Ethereum developers pull off these upgrades and if crypto advocates manage to convince the major powers to legalize the zk-proof cryptographic methods needed to provide blockchain transactions with privacy.


Yet most modern AI tools and any modern technology took a shorter time to reach critical mass.

We are not in the 1960's anymore, 15 years for something that was made in 2008 and isn't used for its intended use case is not early days at all.

It is a failure.


AI, a field that’s almost as old as computer science?


cryptography, a field that's so old that evidence of its start is lost to time, but at least 3900 years.


As a cryptographer, I can confidently tell you that cryptography the science only became a thing in the early 80s. The caesar cipher is irrelevant to modern day crypto.


AI has been in development for decades. Practical artificial neural networks have been in development for 40 years.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: