4 ms·
> If it's published via HTTP, it's mine to consume the way I choose and simply reading it has no legal implications on my part. I don't understand the connecti
by ovao 8y ago
> If it's published via HTTP, it's mine to consume the way I choose and simply reading it has no legal implications on my part.
I don't understand the connection between HTTP and entitlement. Can you explain why you feel that one implies the other?
- cornholio 8y agoIt's the same type of entitlement that makes me free to listen to anything spoken in a public space. We can debate what is the public space in this case, is it a publicly discoverable or password protected URL etc. - case in which is not really "published", i.e made public. If it's legal for me to download it on my machine, then I should be able to read it and disregard any prompt (or setup my agent to so), it's clearly a non-binding contract.
- ovao 8y agoOutline's servers are not a public space. The idea that discoverability by protocol implies ownership or a grant of license seems faulty. With respect to Outline's terms, I'm not a lawyer, so I can't speak to whether they're necessarily legally binding. My guess, based on how they're presented, would be 'no'.
- cornholio 8y agoI'm not sure you fully understood the problem or that we are speaking about the same thing. Outline.com is a paywall bypass service that republishes original material found on sites like the New York Times. I expect this service to be illegal, I said so myself in the first post. Outline's terms are irrelevant here, they are the ones clearly breaking NYT's terms. That being said, I went on a tangent and expressed a political preference that any "Click to continue reading and sell your soul" prompt should be non-binding and functionality to bypass them should be added to browsers via extensions. New York Times servers are not a public space but the Internet is widely recognized as such. A published article trying to force a legal contract as a result of simply downloading and viewing the text, on my private computer, via a public space, should be no more binding than email disclaimers. Again, this is a political position, not legal advice.
- basch 8y agois outline ok if it is client side js that grabs the page and rerenders it? is it any different than a chrome extension. how about if they give me a virtual machine in their datacenter, and i run chrome+client side js, inside my vm? You could get into Aereo vs American Broadcasting Cos territory if your service made the claim that it dynamically rendered the page for each request, using private instances. now lets say you mix in some ipfs+cdn? where do you draw the legal, illegal line? if you really want to get crazy, clients or competing cleanup services (ublock, outline, mercury, instapaper, pocket, user hand cleanup) could compete for the best re-rendering method, one page at a time, and the distributed database of pages+alternateviews could keep track of which "view" is considered the best, for each page.
- ovao 8y agoSorry if I misunderstood. There's certainly a lot of room for discussion and debate about what terms are binding/enforceable and what aren't. Terms of service and automatic user consent are pretty legally wishy-washy concepts in general. I understand the argument you're making, but I think it's not all that dissimilar to the argument a person might make about software piracy: that anything discoverable on the Internet becomes, essentially, fair game because it inevitably touches your private device. From a technical perspective, the argument's correct. From a moral/political perspective, I don't think there's a meaningful distinction between the two.
- cornholio 8y agoThe software analogy is an interesting one. It is generally accepted that software, when ran on a computer, embodies a device or service that is fundamentally different from the same data simply displayed in an editor as computer code. Click-through EULAs are widely recognized and enforced by courts, quite the opposite of email disclaimers. So in a sense, you do have the right to download any computer code you find on the internet, if legally published by it's author. You can print it out and study it's behavior and the author has no way to restrict that since they themselves published the information. But you typically cannot run it or modify it outside the license terms. The distinction hinges on the interpretation we give to the data, and I guess that opens a fatal flaw in my argument: while a HTML version of the article constitutes a document that can be downloaded and read regardless of any disclaimers attached, a binary blob that must be run to deobfuscate the text is in effect a program. And since the hallmark of computers is their ability for general programming, it's quite easy for publishers to force you to run a program and accept it's licensing terms before providing the information you seek.