Even if it is linked externally, Steemit serving the links could be liable for using copyrighted content in their business. Although it is on the Steem blockchain, Steemit doesn't have to serve the content. They should implement a copyright compliance team. Steemit is a centralized entity and thus is liable.
That is unless the user is downloading a client and the user is deciding which content from the blockchain the user wishes to request and view.
So that's why this guy made $40,000 on this one single post.
https://steemit.com/piston/@xeroc/piston-web-first-open-source-steem-gui---searching-for-alpha-testers
Can the blockchain handle every user downloading the content in real-time from the witness nodes
Witness nodes don't serve data to users. They just sign blocks. They're run locked down without any user-facing services.
User data is provided by API servers, similar to Electrum nodes. Anyone can run one. Whether that becomes a paid service (or similarly one provided by app sellers to their users) remains to be seen.
Most of the content is pretty small anyway. As discussed a few comments back, it is just text and links. Embedded media is external. The bandwidth requirements for a full node will be high due to the size of the blockchain, but one such node will be able to serve many, many users with relatively little bandwidth per user.
So as of yet, there is no economic model for serving bandwidth directly from the blockchain, because the serving nodes are not paid.
This is a significant issue because how do you pay them? If you take it from debasement, then how many do you decide to pay? It wouldn't be a market. If they charge users fees, then the white paper needs to remove the lie1 about 0 transaction fees (although these wouldn't be transfer transactions). Besides there are transfer transaction fees when you transfer in STEEM due to the much higher level of debasement while holding STEEM (even for a short interval of time).
The other problem is who pays for the bandwidth for sending the data for witnesses to send to all these serving nodes. If the witness nodes send only to one or two serving nodes, then those have a privileged position. Again who decides how many serving nodes there should be. It isn't a market.
The solution obviously is the users must pay fees to the serving nodes, who then pay the witnesses to serve to them. The witnesses would raise their fees as necessary to limit the number of serving nodes they serve to, in order to maintain their performance attributes.
So if users have to pay fees to use the network, then their SP balance needs to be transferable, which is impossible in the current business model of free signup providing from free SP that is locked up until they earn 30 SP from site activity.
See the insoluble dilemma now?
(Obviously you can tell I've been thinking about how to make a superior clone)
Obviously they can probably just go with adhoc volunteerism short-term, but to really scale to millions of free signups they need to solve this.
The only solution I have been able to think of is that Steemit has to be monetarily incentivized and it must pay for this network serving cost so that users don't need to. But then Steemit, must address copyright compliance. This is the solution I am thinking of implementing for a clone of Steem+Steemit. Note I would replace the dysfunctional voting system (which will prevent Steemit from crossing the chasm away from its circle-jerk blockfantasy demographic) with something that isn't voting and works correctly to incentivize the highest quality content with sub-communities demographic targeting. And of course others can create other serving sites to compete with the original one, but the users will have to pay on those. Eventually all users will have to pay for network bandwidth. There is no way around that but hopefully a revenue model can pay that such as some sublime (inconspicuous and unobtrusive) advertising. The main serving site can be funded by the blockchain during the initial phase of onboarding millions of free signup users.
Specifically I am contemplating a design with no debasement at all even if they are not locked up for 2 years. Much superior for investors. Superior for users also, but I will explain that later.
The irony that I had thought of in 2014 the concept of voting from debasement before (but not the 2 year lockup and STEEM vs. SP split) and dismissed it as dysfunctional (per my explanation yesterday). I had thought of in 2015 to incentivize free signups from Facebook and verifying they are real people. And I had thought of in 2015/16 this onboarding as necessary to create an ecosystem around my microtransactions blockchain design. But what I hadn't thought of until I saw Steemit in action was the ability to pull in new users because of the ease of creating blogging content and rewarding them monetarily. I hadn't thought of making the signed up users into content creators of that sort. I was thinking of them being more of sharers of content created by existing indie artists, e.g. music (which I still think is also a valid onboarding model as well).
Afaics (perhaps I am wrong?), Steemit is a copyright clusterfuck thus far. There are copyrighted images being used all over the site. The copyright holders will come extract blood from the Steemit Inc.
1 That is another lie that I forgot to add to my recent blog posts "Lies about Steem and Steemit", with the word "lie" freaking out AlexGR.