josiahturnq

Ironically, capitalism is all the things you love, which makes the caption a bit misleading. The thing you hate is where the government got involved and F'd up healthcare.

It's week 7 of this and it's getting, candidly, way harder. Making progress is.. - more about game balance than it is creating new features - more about UX design than core content, - more about managing a product than just building my dream game. Which is fine, but the dopamine payoff is way lower so it takes real grit to keep pushing through, especially without a lot of players (we had almost 1k last week, but only a handful are still playing). So anyway, as usual, three parts: (1) player acquisition, (2) what I'm learning about vibe coding, and (3) actual gameplay updates (comments, as always). --- First, the numbers. I made a post on Hackernews (which I have since realized is against their rules) and I had an influx of nearly a thousand players. To my surprise, it was almost all positive feedback. A lot of people thought it was fun, and the only haters were those who hate AI, and didn't have a bone to pick with the game itself. So the stats that came out of that were ~800 players, 240 hrs of playtime, and my top players basically logged 30 hours in just a few days. Unfortunately that's just about ALL my players. So 4-5 power users are still coming back daily and that's it. Which gives me a stickiness percentage of around 2%. Now it's kind of hard for me to understand what to do with those numbers because I'm also advertising exclusively to people who do not care. So, is that high? Low? I have no clue. For anyone that has some actual experience (unlike myself), I would love to hear thoughts, strategies, and and conclusions on whether or not I'm doing OK on the user retention front. If not, it's probably better to call it quits by the 2-month mark. --- Second, the vibe coding experience. I think Anthropic raised the usage limits for the first few days to some ridiculous number, because I got an unbelievable amount of work done before the Fable 5.1 token reset. Since then, it's been very limiting. In many ways, Opus 5 does a great job coding, making art (ish), and everything else. But for communication and simply following commands, it's just painful. I have documented it multiple times doing the literal opposite of what I asked. For example, after very specifically designing a handful of GUIs in Claude design: Handed Opus the design spec I return to check on progress hours later. Opus claimed to have completed it. The design spec was not met at all. I push back and say "this looks nothing like the design!" Opus's verbatim quote: "it isn't a near-miss — it's a different layout. Same for the Forge and Warband ones, which I never opened at all." Like, what? Man, is that frustrating. Honestly I just need to know, has anyone not found Opus 5 to be infuriating? What are you doing right? --- I'd love some feedback for anyone interested. Links below. Game: https://eldermyr.com Discord: https://discord.gg/8MaaAemDBu

I created this because I've been vibecoding a lot and the AI-generated art is good at its core, but with tons of defects. Now we can all edit our art simply and quickly!

Oops wrong link. https://josiah-turnquist.github.io/myrling-sprite-editor/edi...

I created this because I've been vibecoding a lot and the AI-generated art is good at its core, but with tons of defects. Now we can all edit our art simply and quickly!

Ah my bad, thought this was an open forum!

good idea

we're coming for you, Rockstar

Interesting comment but frankly I'm speaking as someone who has no idea how to market, not as an AI salesman. Valid concern, but sounds like confirmation bias.

added the overhead chat display as well!

Chat tab has been added so it's not so easy to miss chats now.

Didn't realize this, that's a massive problem haha. Thanks for the note!

fixed!

the world shaking all the time has been fixed now!

Apologies for anyone that got kicked -- I wasn't ready for the 300 players in 1hr. It does indeed scale with player usage, but wasn't quite ready for 35 players at a time!

Yeah this was a new find for me as well. I've never had enough players for this to become obvious, but I think every player's ultimate is causing every other player's screen to shake... new find, but really annoying. Queueing it up.

This is awesome feedback! I'd love more of this stuff if you have time for it. I have a discord link in the game or just chatting here (if you have the time to spare)

Fair callouts

Fair. I respect that. I'll work on it!

I get it, you don't like AI haha

With 6 weeks in, we're working on the massive part lol.

Frankly nobody will click into something that is just a game after 6 weeks. It's extremely unlikely to be interesting. 6 weeks of vibecoding brings the AI hype, and sets a truer expectation of substantial progress.

Plus all the AI guys are intrigued. I would be, as one of them.

Yeah making game mechanics is the easy part. A game that's actually fun is much harder and I'm struggling with that.

bummer! Just did a performance optimization but haven't tested in firefox. I'll look into it.

There is a storyline with the legion, but generally most users aren't motivated by that. I certainly am not in any games I played growing up.

Noted tho that there are users here for the plot. So a plot they will get!

Would love feedback!

It's week 5 (and a half - like got me) and we made some huge progress.

As usual, I'll break it down into three categories: (1) players acquisition, (2) what I'm learning about vibe coding, and (3) actual gameplay updates (this will be in the comments)

---

First, my Reddit posts are getting less and less attention (not surprising) but actually having way more success on the player-front. I received some really valuable feedback a couple weeks ago from Redditors and someone named BingusDingus (that's his IGN, at least).

Basically their input was that the intro sucked, users were met with a wall of text instead of helpful instructions, and received no context of the story game arc.

So that became my next focus point. And once I put up an interactive tutorial instead of an instruction manual, my user retention went from 35-45s per new users to now sitting around 4min per player (median), and averaging 23min of total playtime per player.

Even more surprising -- I've had 63 players in 10 days, and 5 of them have spent over 2 hours each. I had a great convo with one of them (BingusDingus) but the other fellas are a mystery. Once has players a dozen different times by himself. So, kinda strange, but exciting.

That all said, either I'm getting tired of my game or I've just been staring at it too long. So I would love more feedback on that first 5min and whether or not there's a "hook".

---

Second, the vibe coding experience. I have many friends saying basically the same things - which is that ChatGPT Sol is way better at communicating and designing interfaces but is simply untrustworthy for big tasks. Meanwhile the Reddit communities seem to vent about how awful Fable is and how amazing Sol it. So I'm a bit mystified on whether I should try switching.

Either way, I've switched my strategies a bit and use Fable for a LOT, and only use Opus 5 for very specific tasks, and it's really helped with its ability to stick to a task and actually complete it.

My one massive painpoint is that Fable is GREAT with web design. But holy moley even when using Claude Design, the GUIs are just awful. Has anyone else had experience trying to make a good game GUI? I'm talking like, it's putting raw text over the active screen and so players can't even read it. It's like bad-bad haha.

Oh and one more thing - a strength of Fable's is actually balancing the game. If you give it the variables ahead of time. I found out very much by accident that my archer's had crits stacking for 15x higher than they should've, making them absurdly OP. Turns out I had 7 different skills stacking. It really screwed me when I realized I was using an overpowered hero to balance the game...

---

I'd love more feedback for anyone interested. Links below.

Game: www.eldermyr.com Release Notes: www.eldermyr.com/release