From @dimitarvp on Fri Mar 17 2017 03:18:20 GMT+0000 (UTC)
The title summarizes my worry and I’d like to elaborate my reasons for asking the question.
To clarify: a “paid pin service” is basically a company with IPFS nodes with hefty storage attached that you pay, supply with hashes, and they replicate your content on their nodes.
What I list is personal and subjective preference. I am reasonably convinced by my HackerNews and Reddit frequenting that many other techies have the same preferences as me. Still, we aren’t representative for the entire IPFS community, obviously.
-
IPFS should be used to evade censorship, reduce/eliminate echo chamber effects and make it hard for oppressive regimes (like North Korea) to bribe/coerce a small amount of entities (press, influential bloggers, national TV outlets etc.)
-
IPFS should be used to nullify man-in-the-middle malware attacks. Several NSA and CIA leaks have clearly shown that encryption works. So the spy agencies’ tactic is to plant infected software and bypass the problem altogether. I only heard (sorry, can’t provide links) about some popular software installers being intercepted along the way and modified to include spyware – that includes replacing the MD5 / SHA1 checksum that one can use to verify if the installer is correct as well. IPFS’ very nature directly kills such attacks. It absolutely must have a community-curated big list of hashes pointing at popular software installers.
-
IPFS should automatically ensure content is distributed and accessible everywhere with a reasonable lag – probably making sure it has at least 3-5 copies on other IPFS nodes in the process. I won’t get angry if my IPFS node takes 3 minutes to find seeders for the file I am interested in. Unless IPFS’ goal is to fully replace the WWW (doubtful), I don’t think that’ll ever be an issue.
The point 3 captures my worry about IPFS’ future the most. I am already starting to see paid pin services and IMO that’s severely disappointing and dangerous for IPFS.
In my eyes this must be a baseline feature of IPFS, not a paid add-on service!
The current implementation of IPFS is pretty good and I sure as hell wouldn’t want my node to start distributing my content without me explicitly telling it to do so, yes. But if I issue a publish command of sorts, I’d expect everybody with access to IPFS to be able to find my content several minutes later at the most.
When I first heard of IPFS, I thought to myself “now THERE’s the solution of many of the current Internet’s problems!” – recently though, it feels bad when I scour through the issues and read the discussions. A standard IPFS node implementation should definitely take care to automatically disseminate content if so instructed. And every IPFS node should have a shared host mode: a setting specifying how much gigabytes of your own storage you are willing to donate.
I understand this is much more easily said than done of course. I believe IPFS must aim to be a solid tech with a clear vision and right now I feel that vision is somewhat lacking. It shouldn’t be “good enough”, it should be “the next evolutionary step”. We have enough half-finished technologies as it is.
Don’t the IPFS authors and contributors want to stand out and make a real difference?
Copied from original issue: https://github.com/ipfs/faq/issues/239
