Deltarune – Chapter 3 & 4

Toggle blocks contain spoilers. You have been warned.

I wrote about Deltarune back in 2022, but here we are in 2025 with 2 more chapters that I finished in about 10 hours straight last night.

It’s always hard to know what to say about Deltarune, and by some extension Undertale. As far as games go, I feel like you either enjoy things from a surface level, or go full Pepe Silvia. There is no in-between.

In that sense, then, the points I made three years ago still stand. The music slaps, the writing is great, and the actual mechanics and gameplay have continued to evolve in interesting ways, pushing the mini-game and bullet hell formula even further than before, as has the out of combat traversal.

Absurdity and Sincerity

I’ve been struggling to put my finger on why I feel like other games that imitated Mother and Undertale haven’t quite succeeded. The premier example of this is Knuckle Sandwich, but Athenian Rhapsody runs into some of the same issues.

All of these games trade on absurdity and weirdness as defining aesthetic traits. This can be strange characters, stories, items, or whatever. But at the same time, they’re trying to be heartfelt, sincere, and meaningful. This is a difficult balance to strike. And often the other games fail because in order to do this sort of thing, you need to commit to the bit.

Chapter 3 of Undertale follows the story of a television consumed by fear that it will be thrown away because no one watches it anymore, uses it, or plays games on it. Fear that it’s old and unloved.

Its goal seems to be preventing the player from ever leaving or giving up: from escaping. But it’s a real, human fear given to an inanimate object. Now, in the big picture, it’s complicated why no one is watching the television anymore. But this chapter’s absolutely gonzo section of puzzles, games, and just general weirdness (including fighting a water cooler) is driven by this sincere emotional beat.

And again, we’re talking about a television here.

And I think this is why Deltarune works. Even when its characters aren’t human, their feelings are. The problem with absurdity is that so often it’s used to ignore consequences and responsibility. “LOL random” humor is at some level as disposable as a dream, as transient as a breeze. And that disposability is the opposite of meaning.

For an action to mean something, it must have a consequence. It must have weight. I think this is what Mother and Undertale understand, but that their imitators only grasp for brief moments.

All that said, the other comment I do have is that I think a friend of mine who decided to wait until the full game is out might have made the right choice. Apparently 50% of the game is currently available, and while it’s incredible, it’s also deeply unsatisfying to know I’m going to have to wait at least 6-8 months minimum for more story.

I’m also really hoping the chapter based design of the game doesn’t lead to the final project feeling disconnected and incomplete. Chapter 3 is a massive bit of tonal whiplash, at least in the moment. It’s not a bad thing, I just wouldn’t want 5 more chapters just doing that.

Anyway. Deltarune. Incredible game. Love it. Really hope it finishes development before 2030, the death of democracy, and/or the end of the world.

Some quick thoughts on AI in the workplace

Or
This is the sort of thing I would put on LinkedIn if I cared at all about LinkedIn

I got an email this morning informing me that I was only using 3 of the 4 AI tools that have been made available to my company at the expected rate, and I would need to begin using the fourth, and my mind immediately went to XKCD 2899.

The idea of tracking AI usage actually makes sense to me, but as a metric, not a target. Tracking it as a target implies to me a certain level of buy in, belief in the assumption that this stuff makes you more productive at your job. And I’m not actually convinced that’s true for me.

Information Sourcing and CYA

My primary issue with AI in a professional setting (please don’t confuse it with my personal opinion) is that it does very little that’s actually useful for me. I can think of one general use for the stuff that I use willingly, and it’s a internal RAG bot.

Why do I like this RAG bot? Simple, it lists all it’s sources, and it’s hooked up in such a way that it’s better at searching our documentation (internal and client facing) then any other tool. But I don’t automatically trust it’s summaries.

I’m working on an internal cross-team project at the moment with a fair number of my co-workers, and I was a bit surprised to see them treating information from this bot fairly credulously.

I like these people. I trust their judgement in their areas of expertise.

But I will walk barefeet across broken glass before I quote a price to customer in the seven fucking figure range for a feature because an AI told me it was accurate without double checking that shit first.

And I suppose this is why I don’t like workplace AI much. If it screws up, (and as of 2025, this stuff DOES screw up) its ass doesn’t get fired. Mine does.

And I do not trust a bunch of rushed to market, hype driven, LLM’s with my personal job security.

So going back to that opening point: I’ll use our AI tools. But for most of them, I’ll be using them because I’m required to, not because they solve a business problem or need I have.