The percentage of dofollow to nofollow is not a factor in Google's algo and never will be. It's a myth...as is link acquisition rate, deep link ratios, etc.
Nofollow links do not help a site rank and it's the easiest thing to test if you don't believe me. Google can see where the nofollow link points but it does not pass any weighting through that link.
Link building just needs to be approached as if you were properly marketing a business/website. People get it wrong when they try to game the system. Effective marketing generates links, followed and not followed....it's not that hard.
I disagree.
So are you saying that if a website had thousands of backlinks, and every single one of them was do-follow, Google would just ignore what is an extremely obvious indicator that the backlink profile has been artificially built?
Like you said yourself, effective marketing generates followed and not followed links. They know what looks natural, and can use these factors to determine quite easily whether link profiles are natural or not.
Google gets the best of both worlds with their strategy. Everyone has been put off bothering to build no-follow links, whilst Google can still factor them into their algorithms to help determine relevancy and the genuineness of the overall link profile. I'm sure there are also dozens of sub-algorithms around that which safeguard against various scenarios.
Just look at Interflora when they had their rankings wiped out on Google a few years ago. They paid to have advertorials on popular newspaper websites containing do-follow links, and Google found out within days and slapped a big penalty on them. They don't helplessly rely on no-follow and do-follow tags alone to crawl genuine links. They're quite good at figuring it out on their own these days.
Links are not just about passing juice any more. Google has had 17 years to develop their algorithms with thousands of Ph.D. engineers. I would not be remotely surprised if they factor in absolutely everything they can find on the web, but just utilise certain types of data differently.
It would not be particularly difficult at all to write algorithms which factor in
parts of social media data in specific ways along with safeguards to prevent manipulation. These tests people refer to (i.e. start social media activity and see if the rankings increase) are too simplistic, as Google will have safeguards in place to precisely stop that sort of activity from having an effect.
It's also easy to provide different weights. For example:
Do-follow links on normal sites - x1.0
No-follow links on normal sites - x0.5
Links on social media - x0.1
Tie these three together with a ratio analysis to determine natural or manipulated activity, and you have a system where social media's effect is too small to change rankings on its own, but can be factored into the wider picture to help Google determine quality, relevancy and authority.
If you had dozens of petabytes of internet activity data at your disposal, wouldn't you figure out a way to harness it in some form too?