[Aachen]

Copyright is only part of the equation, there's also the use of other people's resources

If what a government receptionist says is copyright-free, you still can't walk into their office thousands of times per day and ask various questions to learn what human answers are like in order to train your artificial neural network

The amount of scraping that happened in ~2020 as compared to 2024 is orders of magnitude different. Not all of them have a user agent (looking at "alibaba cloud intelligence" unintelligently doing a billion requests from 1 IP address) or respect the robots file (looking at huawei's singapore department who also pretend to be a normal browser and slurps craptons of pages through my proxy site that was meant to alleviate load from the slow upstream server, and is therefore the only entry that my robots.txt denies)

[lolinder]

But here we're talking about Common Crawl being included in this scheme, which is explicitly designed to make it easier to use them than to make your own bad robot.

You block Common Crawl and all you'll be left with is the abusive bots that find workarounds.

[chii]

> you still can't walk into their office thousands of times per day

why not?

Esp. if that receptionist is an automaton, and isn't bothered by you. Of course, if you end up taking more resources and block others from asking as well, then you need to observe some etiquette (aka, throttle etc).

[Aachen]

> why not? Esp. if that receptionist is an automaton, and isn't bothered by you

I chose "thousands" to keep it within the realm of possibility while making it clear that it would bother a human receptionist precisely because humans aren't automatons, making the use of resources very obvious.

If you need an analogy to understand how an automated system could suffer from resources being consumed, perhaps picture a web server and billions of requests using a certain amount of bandwidth and CPU time each. Wait, now we're back to the original scenario!