This is not about the stack. Not FSD, not Supabase, not another PWA. I am 37. Sixteen years in development, web systems, releases, architectures. I like that work. I am still sitting down to my own things, including mobile games on Godot.
Not because I am “tired of the web”. Because I want a thing a person opens on a phone between errands, not only help someone else’s takeoff. The user is no longer a company with a committee. A person with a minute on the metro. That minute should feel good. So should mine: otherwise I get another delivered project with no life in it.
The base is already there: products that live and bring money. The game is an investment, not the only woodpile. Runway is counted. I write this here so there is something to ask me about in three months. A public promise is cheaper than another quiet plan in the notes.
What I am doing
Short cycles: idea, prototype, playtest, a decision from the numbers. In public. Visibility is the best project manager I am willing to hire.
Mobile is meaner than ten years ago. ATT broke cheap performance ads, CPI went up, hyper-casual shrank. Products with retention and a niche live. Educational games still stand on organic traffic and articles. That is the field I am stepping onto, not another three-in-a-row with an ad on the third move. I pick Godot because one person should reach a playtest, not spend half a year assembling an engine. Mobile platforms, short sessions, no cinema at the start.
The game: Memory Palaces
Not “repeat the sequence”. Gentle mnemonics through a game loop.
First, simple rows to catch a rhythm. Then hints: associations, mini-stories, images. Then a visual palace: you place objects in rooms and bring them back in order.
In spirit it sits near Monument Valley for a clean frame, Duolingo for a short lesson, Lumosity for a skill focus. The subject is different: mnemonics themselves.
MVP in 2–3 weeks: about 20 levels, two or three task types (objects, words, numbers), one base palace, progress tracking. The rest after playtests. I do not marry an idea. If it does not fly, the next prototype.
The first 90 days look like this. First a prototype and a warm audience. I watch D1, D7, session time, how people finish levels. Then more palaces, themes and variations, a light frame that gives a reason to return. Then organic traffic, small paid shots, and a decision on the money model.
Numbers I will not skip
Starting targets: D1 from 35%, D7 from 15%, average session 3–5 minutes. Without that the unit will not add up. In the first month I want about a thousand players on a warm audience and to watch completion quality, not only installs.
Channels: short demos on Shorts and TikTok, writing on mnemonics, paid traffic in small shots to test a channel, not to “scale”. Model: freemium plus a subscription for techniques and themes. No paywall on the first breath.
Sleep on a schedule, a demo once a week. Burning out on the first palace is stupid. Living money already sits in other products. The game should not become a theatre of escape from the web.
If it does not fly, that is not the end of the chapter. It is the end of a prototype. The next one will be short too. I am not going to marry the palace idea and polish art for three years. Either the numbers exist by the end of the first month, or the idea goes on the shelf next to the Telegram version of PassWave.
If you want the first beta circle, write. Next to this sits an older note on how a portfolio lies to its owner: Portfolio amnesia.
