ChatGpt is finished training and doesn't update itself in any way.
The current conversation is fed each time as additional context back when you ask additional questions.
If the chat is published on the net, and becomes part of ChatGPT's future training data, I wonder whether ChatGPT could somehow become "aware" of it in a way that would change its subsequent behaviour?
The python code also prints the resulting list in the same order. It's a completely different problem to solve.
In python I would throw the list in a set and return the size of it / print that.
An use case that comes up quite often is building a List / Map / Set implementation. You want to have it to enforce to (typesafe) take only objects of the same type, and when taking it out again, get the object in the same type back. And you want the compiler to complain, when to try to put a String in a List<int> (a list of strings), or try to asign the object you get out of the list to a boolean.
And obviously you don't want to write a specific List implementation for every Type, like a StringList, DoubleList, IntList,... especially as every method in it would basically work the same.
You can't really put content on the blockchain, that costs a surprising amount because of all the proof-of-waste. What happens is people put the content on a host and the URL on the blockchain. I suppose in theory it could be kept in IPFS, but few NFTs seem to bother with that.
The idea that people put the URL in the blockchain is misleading and people keep repeating it as a mantra.
Yes, you put a URL on the blockchain but it's almost always a URL to the ipfs protocol e.g. ipfs://QmVc6zuAneKJzicnJpfrqCH9gSy6bz54JhcypfJYhGUFQu
This identifier is a base58 encoding of a sha256 hash that points to the content. It doesn't actually tell you what host to use to find the content, that's completely left to the client app; so you could use an ipfs gateway like cloudflare-ipfs.com or ipfs.io or you could, imagining a future in which ipfs is not used anymore, use any other kind of p2p network to find the content.
Just like gnutella, or torrents, as long as someone out there is hosting the content it's usually possible to find it.
Except ipfs content disappears if no one is pinning it. And you can find IPs of hosts which are pinning it. Yes there is FileCoin but there are minimum and maximum file size and contract duration limits. So someone has to periodically check pins, contracts and keep purchasing new FileCoin contracts to keep it online.
Yes, true, I don't think ipfs is meant to be censorship-resistant though, so finding the IP of hosts doesn't seem like a big deal for the project goals.
My point was more that the NFT metadata itself could still be usable x years in the future when another protocol has replaced ipfs – it would just have to make the files findable by the existing hash.
The biggest potential risk would be if sha256 becomes broken and finding collisions is trivial.
What blockchain? Some newer ones are not expensive (but also are not decentralized enough yet). They dismiss ipfs straight away in the responses already for it and actually say they will put the actual imagine on the blockchain. This will cost something (possibly a lot) depending on which chain, but they have $4m+ left to burn on ‘the good cause’. This all stinks but I must agree, if you want something for eternity (well, in their minds), putting the content on a big (btc/eth) chain would be the only way; seems the rest is far more brittle.
Jikes, that's even worse than I thought. So then they will probably just go for another coin and store it there. Not sure how else. Or maybe they'll raise a few billion after this victory /s
Unfortunately, things like that actually happen even though everyone here would consider it a joke.
A handful of words can be copyrightable. The idea of copyrighted content being stored on a blockchain that thousands of nodes duplicate and offer freely is an interesting conundrum that your reply entirely sidesteps. I’m guessing it has happened already but I too am interested in the legal implications of this.
Legally it's not complicated: is it a copy? Is it permitted by the copyright owner? No? Then it's an infringing copy.
Actually doing something about it is a problem, as shown throughout the post-Napster era, but every node that serves an infringing copy is potentially liable.
* Is the material copyrightable? A novel is copyrightable, but a recipe is not. The layout format of a telephone book is copyrightable, but the contents of it are not.
* Is the copyright still in effect? Unfortunately, this is a less relevant question now, as anything written within living memory is under copyright, but it is still a required step.
* Is permission from the copyright holder required? If I am sold a copy of a computer program, I'm allowed to make whatever copies are necessary (e.g. copying from the hard drive to RAM) that are necessary to execute the program, without the permission of the copyright holder.
* What is being done with the copy? If I buy a work, the copyright owner has no right to prevent me from selling it, nor are they entitled to any portion of the proceeds. Re-sale of authorized copies, regardless of permission, are not infringements.
Well, they won't. All NFTs are hosted on hilariously centralized services like googleusercontent.com and similar.
Browse through https://opensea.io/, the most prominent NFT marketplace at the moment. Digging through the DOM of their website will reveal a lot of links to quite standard content storage solutions. They usually go through a bit of trouble to prevent people from simply clicking "view image in new tab" to reveal the actual location, but it's not difficult to get around.
Almost all NFTs are hosted on IPFS. Opensea shows a CACHED version of the image, because pulling them directly from IPFS puts unnecessary strain on the network and is comparatively slow.
The contract for the NFT generally just points to an IPFS identifier (a sha256 hash of the content), and it's up to the viewer (a website like opensea) to decide what ipfs gateway to use. Even if the ipfs network died you could still use other p2p methods to find the content, as long as someone was still hosting it.
The amount of artwork NFTs that has "Metadata: Centralized" in their details appear very high, although it's difficult to get any exact numbers. As far as I can tell, OpenSea doesn't provide an option to filter based on this field.
Some high value collections like BAYC have frozen metadata, so it can never be changed. I do know some other big collections serve images from AWS or their own domains, this happens because it's extremely expensive at current gas prices to do partial reveals or any kind of metadata changes on-chain
It's a problem, it used to be the grand majority of collections were fully decentralised and frozen but nowadays it's a mixed bag – I think it will change once the gas problem is solved, but it's also partly because the audience has changed and newcomers into NFTs don't care so much about their tokens being fully decentralised.
Did you actually read the post? They propose to put it on-chain, and even say
> [O]ur NFT collection could very possibly become the largest on-chain collection ever in the history of Ethereum. Time would only be on our side because as gas becomes more expensive, the window of opportunity to bypass our on-chain NFT collection would fade away.
Remember that this is from the people that believe buying a book gives them the IP rights to the work (should have bought a Harry Potter book). They may not have thought it all the way through.
To give an idea of the storage costs associated, I dug up this old stackexchange-answer about storage costs on Ethereum [0], which estimates that it costs about 76.000 USD / GB stored. Note that this is 5 year old answer, so the price will probably have multiplied like the price of ETH since then.
Edit: Here's [1] a newer, updated price estimate that takes current prices into account : 309.9 Million / GB. LoL!
> Here's what happens when you add a file to IPFS — whether you're storing that file on your own local node or one operated by a pinning service or IPFS-enabled app.
Notice that wording: “Add a file to IPFS”.
It’s likewise fine to say, then, that you “put something on ipfs”.
That's odd, since the TP in HTTP stands for transfer protocol, but the FS in IPFS stands for file system. You'd expect to be able to store data on a file system. That's what I use my file system for anyway.
They are great! My bachelor thesis was implementing an algorithm to generate minimal petri nets based from an infinite partial language, described in something like a basic Regex.
That's also why it's usually a multi-year project to introduce SAP systems at an existing company. The hard part is not installing the software (though I'm not saying it's always easy either). The hard part is getting all layers of the company to change away from their pre-existing, sometimes nonsensical processes.
(Disclosure: I work at SAP, though not on customer-facing software.)
[0] https://de.wikipedia.org/wiki/Rhabarberkuchen#/media/Datei:R...