You have one main point that you are trying to argue for, it seems to me, and I understand what it is. In fact, our disagreement really narrows down to this:

> And I'd disagree that history is too complex to find patterns.

I think it depends on what you mean by “patterns.” if you mean patterns in the sense that physical theories detect patterns that observe formulaic relationships, then I strongly disagree. But of course human history involves human beings. We come with a fairly standard (in a statistical sense) set of equipment. Of course there will be similarities. Just not ones that can be used to generalize across all history.

However, that aside, I wanted to point out that you were actually doing three different things“ below, though arguing for one position:

> If you have all the details, you can loose the ability to use that data.

> Have you seen all of the books on WW2 or Civil War, do you need to read them all and know all details before you can make any judgments, or form any opinions.

> Do you need to know what was going through the minds of every civil war solder at Gettysburg, every movement, to know who won?

First, it’s not really clear what it means to use “data” about history. If you mean make predictions, I think you’re doomed, but that’s because the predictive enterprise itself is doomed when it comes to historical understanding.

Second, I would not consider forming a judgment or opinion about something in history to be in the same class as lumping. So that’s orthogonal to what we’re focused on here. Now, it is the case that, as you split more, dive more into details, your judgments may change. They become more refined and less certain. But part of learning history just is forming judgments about things that have happened and then refining them. The danger of lumping, of course is that you stop refining because you’re so focused on the abstract patterns you think you detect. And the real value of learning history is just in the refining!

Third, who won is a data point itself, so no, you don’t even need to know how Gettysburg unfolded at all to know who won.

Sorry. It seemed you were using Lumping as meaning combining data into larger summarizations, and are against that, or think it leads to errors. Maybe not.

Sure, if there was time to study all of history, that would be great. But sometimes you need summaries to provide context around the subject being deep dived, or more finely split. It isn't possible to split ad infinitum.

In the end all history is summary and interpretations. Some people take it too far, which I think you said, it is a sliding scale. It goes back and forth doesn't it, some people split and split, and find a lot of details, and then wrap it up and write a summary.

For formulas. I tend to think the understanding of the human animal will be cracked eventually. Being complex doesn't mean it cant be done. And we'll be able to find more 'formulaic' models of groups, and perhaps apply to history. But of course, it is speculation right now.

Of course, this 'speculative' history has been going on for hundreds of years. Aren't there historians just to study past historians that themselves tried to find 'models' of history. I'd say lumping has been part of the process for a long time.

But to come full circle, you need details from which to lump, so some groups go out finding details, and others try to put them back together again. It isn't all facts, it is to find out why things happened, and really that takes putting things back together.

Of course we are talking about a couple made up words lump/split, and really each encompass entire fields. Perhaps it is bad example since the linked article lumps too much? So I'm in position of defending lumping, when it started with a not very good example.