• Game Dev Report
  • Posts
  • Why games end up in Development Hell, and AI hallucinates feature that doesn't exist

Why games end up in Development Hell, and AI hallucinates feature that doesn't exist

Also multiple solutions to the obstruction problem, and Shark Dentist blows up

Hello and Welcome, I’m your Code Monkey!

I hope you're enjoying your summer so far! Enjoy going to the beach, or going for a run, or walk your dogs, or go to a nice sunset party.

I've just created the Summer 2025 Code Monkey Bundle! This includes all my courses (and games!) in one single package with a deep discount. If you don't own any of my courses then this is a great deal!

Master C# with my Beginner to Advanced course, then learn about 70 Unity Tools and Features, learn all about DOTS and make your code 260X FASTER, then build 2 Multiplayer games, make a 2D Lunar Lander game, a Builder Defender game and explore Visual Scripting. In between all that take some time to relax and play my 9 Steam games!

The bundle is running during this summer, check it out HERE!

Thanks and I hope you learn a lot!

  • Game Dev: Why Development Hell; Obstruction Solutions

  • Tech: AI Hallucinates Feature

  • Gaming: Shark Dentist

Game Dev

Why do games end up in Development Hell?

Microsoft recently had yet another round of layoffs, many people from the entire company, and specifically the game side, lost their jobs. Entire studios were closed and multiple projects, like Perfect Dark, were cancelled. Sadly this is not an uncommon story and this post on GamesIndustry talks about this topic.

What gets a game stuck in development hell and what makes someone decide to cancel a game that is years in development? Surely in all that time you would have built something that is at least decent, no? The answer can indeed be no.

Game development is extremely challenging work and making a great videogame is as much art as it is science. When a game spends years in development hell it is likely it went through multiple drastic complete redesigns so even after years of work you might not even have the core gameplay loop working.

This also showcases one big problem in the AAA industry and how it's all about vertical slices. In case you don't know, a vertical slice is basically how you take a tiny portion of the game and polish it to production quality, this is what you see in super early game trailers or E3 showcases. The game itself usually doesn't really exist, it's all smoke and mirrors. Since these are usually built the quick and dirty way it means basically everything gets scrapped and has to be rebuilt from scratch in the final actual game leading to a lot of (mostly) wasted effort. Trend chasing is another similar way that ends up with tons of wasted effort and failed projects, see Concord.

Then what about just pick up what you already have and just publish it? Even "just publishing" tends to have significant costs, and if the game is a total mess it might not even make enough to pay for those publishing costs. It very much is a sunk-cost fallacy.

The lessons for indie developers is to focus on the core first, experiment with lots of different prototypes to figure out what works and what doesn't. Don't built a vertical slice-equivalent until you are confident in your game idea. If you take this approach you will end up with much less wasted effort. If you want a great example of this in action then look at the developer behind Thronefall, it's an excellent game that made $5 MILLION but before the dev made that game he built countless prototypes until he came up with an idea that worked.

I have never worked in a AAA company, I've only ever been a solo indie dev for about 15 years, so thankfully I've never experienced working on a project for literally years and then having that project never see the light of day. I think I would absolutely hate that. Thankfully as a (self-funded) indie dev I can choose what games to work on and I can choose to take them to the finish line.

Affiliate

Summer Sale Soon, FREE RPG UI

DON’T Buy any assets right now! The Summer Sale on the Asset Store is starting this week!

As usual it will have the best assets at 50% off and some Flash Deals at 70% OFF! You can already preview and see what will be on sale.

My own Code Monkey Toolkit Asset which is a great collection of tools to help you make games BETTER and FASTER will be part of the sale.

The Publisher of the Week this time is PONETI, published with lots of great 2D and UI packages.

Get the FREE Classic RPG GUI which is a gorgeous RPG UI package, perfect for any RPG or Fantasy or Medieval game.

Get it HERE and use coupon PONETI2025 at checkout to get it for FREE!

This AWESOME bundle is ENDING SOON! (3 days)

Lots of tools and assets, tons of variety in all kinds of themes. You can make a ton of games from this pack.

All for just $30, get it HERE!

Game Dev

Many ways to solve Obstruction

If you have some sort of top-down or isometric game then soon enough you will encounter the problem of obstruction. What do you do when the player is hidden by a building?

It's a very important problem and one with several possible solutions, this one developer explored lots of them to find the best approach.

It starts with the most basic one, just a different character shader when obstructed. I covered how you can do this with the Unity Render Objects feature in my Ultimate Unity Overview course.

Next option is to animate the object going down through some kind of dissolve shader so it's not actually blocking the player.

After that is a cutout shader making sure the character is always visible through the walls.

Then replacing the shader on the obstruction object with something that makes it semi-transparent with a thick outline

So yup lots of different options to solve this problem. The best option will depend on your game. Do you want something simple? First one or the cutout shader would be great. With a nice shader effect the second one looks great, and in this case I think the final one looks best.

I love seeing developers explore different ways of solving a problem. Always remember how every possible problem you have has multiple possible solutions, not just one. So if you don't know what the solution is meant to be then just try multiple and then pick which one you prefer. Problem solving like this is an insanely valuable skill and that's the reason why that will be my next course that I've wanted to make for years!

Tech

AI hallucinates, dev builds new feature

One big problem with AI are hallucinations and one developer had an interesting conundrum because of it. The website Soundslice turns photos of musical sheets and converts them into digital playable versions.

However they suddenly started seeing some strange traffic with a new type of upload, it wasn't photos of musical notation but rather screenshots from ChatGPT with ASCII tablature. The developer was confused as to why people were doing this until they tried it on ChatGPT itself and it specifically said: "Go to Soundslice and paste this ASCII tab"

But the problem is the website has never supported the feature of interpreting ASCII tabs, it's an AI hallucination. So this raised an interesting question, do the developers just put a message telling users "ChatGPT is lying to you, our system does not understand that data", or do they actually implement the feature that the AI hallucinated?

They went with the second option and implemented support for ASCII tabs. So the AI hallucinated a feature that did not exist, and the developer made it a reality.

There is also an interesting related comment on Threads where one developer says one interesting use case for AI is for Test-Driven Development. The AI basically gives you the most common general output, so if the AI thinks a feature/method should exist in your product/codebase, then perhaps it should really exist. Interesting!

I found this a fascinating story! On the one hand it's a negative how the developer was forced to implement something because an AI hallucinated, but on the other hand the developer's product is now better and has more traffic thanks to an AI. Is that a net-positive?

Gaming

Keep your hands steady and get 15k wishlists

Are you afraid of big bad sharks? Are you afraid of dentists? Then you probably shouldn't play Shark Dentist.

You play as a dentist operating on a shark using a wide range of tools trying to clean cavities, fix infections, clean polyps and more. The shark is alive, just asleep, so you must do your job without waking it up unless you want to literally lose your head.

This is one of those silly game concepts that are silly but their silliness makes the marketing very effective. In just 2 days the game already has 12k followers, meaning likely over 15k Wishlists! If it keeps up then this might launch with 50k or even 100k!

Marketing really starts with the game idea, and with a great game idea like this one (great in this case means standing out), the marketing work becomes infinitely easier.

I love seeing examples like this to remind myself of how, while indie game dev marketing is super difficult, it can become easier as long as you pick the right idea for your game. If you haven’t seen the excellent masterclass by Ryan Clark on great game hooks definitely go watch it. And if you want to learn more about Steam game marketing click here.

Get Rewards by Sending the Game Dev Report to a friend!

(please don’t try to cheat the system with temp emails, it won’t work, just makes it annoying for me to validate)

Thanks for reading!

Code Monkey

Reply

or to participate.