General Discussion

You're saying "manipulating results was openly discussed" as if it was a conspiracy to achieve a specific result.

That is not what the poster is explaining. The poster is trying to tell you that statistical methods involve mathematical models that strive to find accurate measurements. In polling, it's voter intent. But everything under the sun is measured in this way.

In some elections, including 2016, pollsters couldn't get an accurate read on how many Trump voters were out there. Trump voters kept getting undercounted. There are a variety of reasons for this. Social stigma, hostility to media or the pollsters, etc.

So they realized their models were lacking accuracy. As a result, they went in and saw what in the model and method could be tweaked that would give results closer to what the reality turned out to be.

You want whatever model you build to reflect empirical evidence. Just like if you make a model about how stars work, then you see a star that isn't operating the way your model says it should, you go back and look at the model. See what changes you can make to account for why yours isn't working. Or maybe there's something about the star you missed that's causing it to behave that way.

It's basic science. No conspiracy. No nefariousness.