Knowing the rules

And breaking them anyways.

Photo by Ian Barsby on Unsplash

I was recently engaged in a debate with a good friend (a different one thank you, I do have more than one) regarding whether I was conflict seeking or avoidant. The back and forth is less important than what came from it. We both agreed that, regardless of whether I sought or avoided conflict, I was principled and that my principles frequently are the impetus for me being involved in some form of conflict.

One of Ernest Hemingway’s quotes on writing that I’ve applied to numerous other fields was to ‘know the rules before you break them.’ His staccato style with frequent punctuation became his signature. An interesting observation I’ve made is that one need not know the rules to enjoy the effort. For most, they’ll never take the time to analyze the works in the way Mr. Hagopian does of Hemingway and Faulkner. Their style and fashion of writing, as much as their content, were akin to a system of expression. These systems are ones we are seeing today with Ai. The emm dash, something I’ve long been a user and admirer of, given my verbose and sometimes cryptic style - are often quite helpful for me and my readers. The same is true for the ellipses.

Sentences should be atomic and therefore punctuated appropriately. These should be combined into a coherent set, introducing, expanding, and closing a thought in a paragraph. The paragraph should open, or close, the adjacent paragraphs. Except… Is that how you talk? Is that how your mind processes information? Are you properly segmenting those items into your working memory of 3-5 ‘slots’ so that you can then pass those on or discard them completely while you work on the subsequent information? No. You don’t. You build a context and you add and subtract from that context as you ingest the words. Your brain busily links them into experiences, information, or frames-of-reference that you already have. If it is entirely new information, you may stumble even on the simplest of ideas.

Interestingly, this is how Ai LLM models work and is why the newest ones are so damn good. They’ve expanded the context windows, and are able to link what they refer to as tokens (which are really just segmented words) into extremely long buffers that give the appearance of intelligence, where none actually exists. Quite likely you too have read that paper, that essay, or that story which seems to miss continuity and flow. It reads not as sequence of intentional moves, but more a random chain of occurrences. Ask any eighth to tenth grade English teacher - they’ll have plenty of examples for you.

So what’s the point to this? I know the rules on Substack now. The algorithm, and likely most of the TAM for Substack seem to prefer video. Their suggestion systems prioritize video messages. I’ve tried it a few times, learned a bit about the Ai generation tools, and burned up an entire month’s worth of credit on an unfinished video. In the end, I’m unimpressed.

I’ll stick with my conversational toned, emm dash and ellipses laden prose sprinkled with a hint of profanity. Thanks for reading.