I know there were other samples posted before that had more of what you are talking about. But I’m strictly speaking about the current batch and the article in question and screenshots from that article.

I’d imagine Facebook would cherry pick the most offending pieces to show off. And whatever evidence is currently presented seems weak at best.

I'd expect the 10^7 follower account to indeed have more influence, but not necessarily be able to effectively use that influence as well as the 10^4 follower account.

The big follower accounts would probably tend to be more general, with more diversity among their followers. For example they might have followers spread throughout the nation. Much of their influence will be wasted influencing people in areas where changing a few opinions is not enough to make a difference.

The smaller accounts might be able to be more focused on one particular region and one particular group of potential voters in that region. They will align with that group better than the big, national account will, and so might be more likely to change positions of their followers, and they can be focused on a region where things are close.

So what you do is use your big accounts as sources, and use your small accounts to post things from the big accounts. The link to the big accounts lends credibility to the material in the eyes of your followers on the smaller accounts.

They could have had tiers of accounts, all for different purposes. They could have used a ton of low accounts to simulate a grassroots response, to openly troll and be banned, and all kinds of things. They’d have cultivated more and less popular, and obvious accounts as well. If they were smart they’d have a whole layer of accounts designed just to be caught, and give a misleading impression of their competence and methodology.