User will need to download wrong object to see what’s inside.
It may be also larger than expected.
Sure, but that’s true on the clearnet too. It would be good if the SAFE Network had a way of showing filesize without downloading, though.
I don’t completely understand what are you talking about.
If you mean that just asking random node to execute code is a bad idea, then yes, of course.
At least several nodes should execute the same code and results should be compared by the network.
But, maybe, even different programming language should be invented to make reliable distributed computations possible.
I’m just saying that’s not a simple thing.
I guess the way I would do it is to reserve a specific namespace for “X data as processed by Y algorithm”. Nodes would have to be designed to refuse to acknowledge and PUTs to that space that haven’t been verified as such. I’m not saying it couldn’t be done, but this seems like something we shouldn’t expect immediately after launch.
I suppose if it’s a decentralized app, the uploader would be the one paying for this processing? And if a lot of people are uploading at once there could be collision issues. Again, not saying it can’t be done, just that it’s not a simple task.
Without distributed computing, developers will be forced to use traditional servers.
I still see that as a win if those servers are limited to data processing. Rather than needing a server accessible 24/7 in a datacenter, you could connect your home computer to the network once a day to process the day’s input.
EDIT: And also, this could be set up to where the “server” software is available for download and can be run by anyone if the original goes offline.