What good is a model that doesn't predict anything?

 So, here's a beef and what I have against 538's model this year, and all of these models that try to predict the actual outcome rather than just the outcome based on what we know now, is that predicting the future 4 months out is impossible. There's too many variables that can play into things that if you actually try to make a model based on it, you get exactly what 538's model does: consistently give both candidates a 50/50 chance because no one actually knows what will happen and what changes.

My model might be relatively certain, and relatively volatile with its certainty, but that's the thing. I'm basing this on the idea of "if the election were held today, what would probably happen?" It actually tells you something and does clearly take a side when the data seems to obviously lean one way. Can my model shift back the other way? Sure. Of course it can. And I will update my model when it does. Just as 538 will update theirs as they get closer to election day. 

I also don't get the anti polls mentality of a lot of people. Well, I do, it's selective reality denial. They dont like what the polls tell them so they're denying what they say and assuming things will change. I'm not sure they will. Sure, they CAN, and I will update my model if and when that happens. But Im not counting on it. I'm taking things as they go and will change my assumptions when the data changes. I'm not trying to predict the final outcome four months out per se. Rather, I'm saying "this is where things are going and what people are thinking and these are the odds as they stand now." 

Again, my model is more volatile, but more honest. I honestly think these fancy models "the experts" use are just sophistry for pseudo intellectuals to use to appear smart, but who aren't smart. You know, like the blue maga pseudo intellectual types who think they're so smart for being centrist dems and thumb their nose both at the right and progressive lefties like me.

Btw, what makes me so firm in my own conclusions? Well, I studied political science, and studied research methods, statistics, and polling as part of my curriculum both in college and grad school. I feel confident in my own conclusions, and also understand my own limitations. My model might not be as sophisticated as others; its on par with 538's old "lite" polls only model. But I do this for fun and because I want to, and I trust my abilities to approximate election outcomes. And I haven't seen these other models actually do a better job.