itch.io

I did not seem to be able to play this game at all, unfortunately. 

I think the audio did not fit the gameplay, the visuals were not enjoyable and often got in the way of actual gameplay, and I could not figure out how to shoot

I didn't really understand the gameplay, unfortunately. 

ETA is about 2 weeks lmao, finished 10 million this time yesterday in just 4 days, you just missed out! So in 25 hours 5 million clicks have been made 🫠 Here's the replay: https://million.wayfarer.games/replay

I can do! Give me a couple of days :)

If I had outright said it was banned, and told people when they were banned, they would attempt to get around it! By doing it this way, people assumed I wouldn't have any autoclick detection so only did it the naive way. I'm confident that the top 5 scores were all human clicks, which means the whole thing worked as intended!

I'll run some tests on the GUI issues, yeah, I'm really curious too 😂

Ah I think you probably just missed it roll over to zero 😅 there's a replay here: https://million.wayfarer.games/replay

There is auto clicker detection, it measures consistency over time, so it'll get them eventually, and remove them from the leaderboard :)

Ahahahahaha thank you so much for playing! I think I am now the most hated person in this jam ;)

That's not the anticheat, it is literally because people were hurting themselves 😅 had to go heavy handed with it and force breaks after two combos!

It's a clicker game where everyone clicks the same number down!

It's not! Everyone in the top 12 is confirmed human 🫠 the autoclickers are on 500k+

I wondered if someone using an auto clicker would show up! I've been keeping an eye on yours, I've got you in a moderation view 😂 you won't see anyone else that high because your score is not that high! You're not on the leaderboard 😉 your score was caught at around 290k and you were knocked down to 50k and stopped contributing to both the main countdown and the leaderboard! Everything appears to continue to work to stop people figuring out they were banned and changing their behaviour lol, it has been very effective.

Thank you for the stress test 🙏 I suspect a lot of the issues you faced (except the audio and gui) were actually because you were banned. You'll be forever immortalised in the ban list ♥️  I can't change the end stats to include you now unfortunately, but I would if I could!

You'll have to wait and see!

The backend is mostly a single script (Lua in Valkey, driven by a Phoenix server) that processes every click one at a time. That means validating the round, rate-limiting, and decrementing the countdown, so there's exactly one winner and no lost or double-counted clicks! There's a bunch of extra stuff around stats, username moderation/autoclicker detection, that sort of thing, but the bulk of it is that one script. I wrote a full post about it here: https://wayfarer-games.com/blog/one-button-mmo/

No no I agree completely 😂 it was a fun technical challenge that seems to have taken off, I cannot explain it at all 🫠

ahaha I see, no worries 😂

sorry what do you mean by "using the rating timer as a countdown"? It started from 10,000,000, everyone is contributing to reducing the same number - every click takes 1 (or 2 if you're on a combo) away from the number. Ratings has nothing to do with it :)

There were a few test ones before the real ones started! Congrats on winning it though, once the jam is over I'll go back through the data and award little badges based on contributions :)

Here's a post explaining the tech 😁 https://wayfarer-games.com/blog/one-button-mmo/

Agree about it's questionable game status, but people seem to be having fun and  it does have a goal, so maybe it counts!

I have several ways, and shadowban them so they don't know they have been detected and need to change their behaviour 😂

It is just a visual readout of your clicking! It's a pretty picture that you make with clicks 😁 partly also a way to detect autoclickers

The top three are legit, I personally know one of them and I've done about as much pattern analysis on it as it is possible to do. They check out! There's a bunch of cheat detection;

The finger break is just for humans! Autoclickers are shadowbanned so they never know 😂

Some people have way too much time 😂 I've done as much pattern analysis as it is possible to do, I've checked out their clickprint, but the top three are completely legit

I wrote a post about it: https://wayfarer-games.com/blog/one-button-mmo/

I wrote about it here 😁 https://wayfarer-games.com/blog/one-button-mmo/

Nah, 10 million is probably too low actually 😂 it will be done by this time tomorrow, no proble

Hey! I made Ten Million Clicks Together for the GMTK jam, a game where everyone clicks one button that counts down from 10,000,000 to zero.

That's the whole pitch. There's one button. You press it. The number gets smaller.

This sounds like it should be the easiest game I've ever made 😂

It is not.

I'll let you in on a secret: "multiplayer" really means "arguments about what order things happened in". A normal clicker keeps its number on your device and updates it the instant you press the button. A shared clicker has to answer a much nastier question: what does "the number" even mean when thousands of people are changing it at once, all with different connections, all seeing a slightly different version of the game?

Nearly every decision in this project is a tradeoff between three things: being correct, feeling instant, and not melting the server, and that's usually a "pick two" kind of situation...

Prediction!

The server has to own the final number, obviously. If every browser got to decide what counted, cheating would be hilariously easy!

Unfortunately, a round trip to the server takes time:

press → send click → server updates → response returns → update screen 

Even a good 80ms connection makes that feel a bit sticky, and on a bad connection it feels completely broken.

The fix is client-side prediction - the same trick shooters use to move your character before the server agrees. When you click, your "contribution" number and combo meter update immediately, and the server catches up a moment later.

But you can only predict things you're allowed to be temporarily wrong about:

• Your own state can be predicted. Your browser knows which clicks it has sent but haven't been confirmed yet, so it can include them early.
• Shared state can't. Your browser has no idea how many clicks everyone else has just made, so the countdown and leaderboard always show what the server said.

I also smooth the displayed countdown towards each new server value instead of snapping it. If 40 other players clicked between updates, the number rolls down through the difference, which feels snappier and more like "individual clicks are updating this number". They very much are not.

The combo meter follows the same rule. The game can predict that the meter is full, but the ×2 multiplier doesn't switch on until the server confirms it. That means there is a tiny delay at the exact moment the multiplier activates, so you never see a ×2 contribution that later gets corrected back down to ×1. In practice though, you're never going to notice this!

You Can't Shard One Number

Most scaling advice boils down to "divide the work between machines". That doesn't really work here, because we have one value that every single player is touching. The countdown itself can't be split into ten independent counters, because then which counter gets the final click? How do all ten agree the round is over?

So I made scoring a single long, messy script that runs on a single thread and is the "source of truth". Every click runs through one small all-or-nothing operation that validates the round, applies the rate limits, works out the multiplier, reduces the countdown, updates your contribution, moves your leaderboard position and bumps a version number.

Usually, you'd never want to do that. I'm sure someone reading this just threw up a little. It does actually work here, though, because the operation is tiny, lives entirely in memory, and never waits on anything halfway through. That means every accepted click has exactly one place in the order!

Batching!

Sending one network message per click is wasteful - people can click fast, and every message carries the same protocol overhead. But waiting around to build big batches is worse, because it increases the amount of unconfirmed stuff on your screen.

So we compromise: the client waits a tiny window and sends up to 16 clicks in one compact message. At normal speeds, most batches contain just one click. During a frenzy, it sends a few clicks per message. The delay is too short to actually feel, but it removes a bunch of work that happens when the system is busiest.

A Timeout Doesn't Mean It Failed

Batching creates another problem: networks are really annoying.

What happens if you send a click, the server applies it, and the reply gets lost on the way back? Your browser only sees a timeout. If you send the click again... have you now contributed once or twice?

You can't fix that with "retry and hope". Every click gets a sequence number, and the server remembers the last one it accepted from you. Batches look like this:

stream: abc123 first sequence: 481 click count: 6 

That means "clicks 481 to 486". If the response disappears, the client sends the exact same batch again. The server spots that it has already seen those numbers and returns the original result instead of scoring them twice.

There are a couple of extra wrinkles. If the browser reconnects and finds the server further ahead than expected, it uses the server's number and carries on. And if the round ended while the clicks were travelling, they are thrown away.

The alternative is choosing between lost clicks and duplicated clicks every time a response goes missing, but I'm pretty sure that would be bad in this sort of game 🤔

Zero Is a Race Condition

The final click is the most interesting problem. When I say interesting I do of course mean "aaaaaaaaaaaaaaaaa[...]"

When the countdown is at 1 and two players click at almost exactly the same time, you might think to do this:

Player A reads 1 Player B reads 1 Player A writes 0 and wins Player B writes 0 and also wins 

Congratulations, we now have two winners 🫠

Instead, that one all-or-nothing scoring operation handles every part of the process, so no other click can sneak into the middle. By the time the next player's click hits the code, the round is already over. A nice, simple, if messy, fix.

Broadcasting!

The obvious realtime implementation is to broadcast the new countdown to everyone after every accepted click.

That works brilliantly with three players and becomes increasingly stupid as the game gets popular. If 2,000 clicks land in one second, sending 2,000 updates to every connected player doesn't make anything smoother - browsers can't meaningfully display 2,000 intermediate states anyway. They'd just burn CPU decoding messages and rewriting the same number over and over.

The trick is noticing that different information has different urgency:

• Your own click comes back immediately, because it resolves the click you just made.
• The global countdown updates are merged together and broadcast up to 20 times per second.
• The leaderboard updates up to 5 times per second.
• A full snapshot goes out periodically, so one missed message can't leave anyone permanently wrong.

The high-frequency messages themselves are fixed-size binary frames rather than JSON, so I'm not shipping the words "remaining" and "players" thousands of times a second. The entire global state fits in 39 bytes.

Yes, that means spectators can be up to 50ms behind the newest countdown, and the leaderboard up to 200ms behind. Nobody can tell, and neither delay changes a scoring decision, because those displays are never used as authority. "Realtime" doesn't mean "send everything instantly" - it means each part of the screen is fresh enough for what it's for.

Fairness!

Rate limiting one player is easy. It stops a single browser from spamming thousands of clicks per second, but it does nothing about someone opening fifty tabs with fifty identities.

The obvious next step is a limit per network - and that's a trap. Hundreds of legitimate players at a school, office or event can all appear to come from the same place. A strict per-network limit treats a successful real-world crowd exactly like an attack, which is a terrible reward for going viral in a classroom.

So the network allowance stretches: the more distinct, recently active players on a connection, the more capacity it gets, up to a hard ceiling. One person can't multiply their throughput by opening more tabs, but a genuine crowd sharing a connection gets room to play.

The Whole Game Is One Number

We have one input and one objective, and that simplicity exposes every tiny little flaw. There's no room for the shared number to be approximately right - everyone has to agree that one ordered sequence of changes happened, even when messages get delayed, duplicated and lost along the way.

Every decision here comes from asking the same question: does being wrong here change the result?

yeah it's not easy, and this is worst-case scenario for networking! I wrote a blog post if you're interested :) https://wayfarer-games.com/blog/one-button-mmo/

One of the major inspirations, and the popcat team actually went to university in the same city as me!

It didn't actually take as long as you'd think 😂 here's what's going on under the hood: https://wayfarer-games.com/blog/one-button-mmo/

I wrote a blog post about it! https://wayfarer-games.com/blog/one-button-mmo/

2x is the most it goes to! Yeah, I wanted something other than pure 1 click counts it down by one, I think that would've been too boring 🫠 

I did write a bit about the underlying tech if you're interested :) the tl;dr is that lots of people changing the same number is a difficult problem to solve, but actually the solution is fairly simple! (just ignore all the usual advice and do everything in one single function)

https://wayfarer-games.com/blog/one-button-mmo/

is this the famous LZD?! How are your hands my friend

This was not a game I expected people to play for more than a few minutes 😂

Do I need longer enforced breaks? 😂 Congratulations!

Thank you 😁 we absolutely will get there! ETA is roughly three days, but that is going to drop HARD when the play & rate period ends I think 😅 I'm pretty sure it will be done before monday morning UK time, and I'll have to make the next one 100 million...

There are three levels of ban for autoclicker use, plus one for usernames - try and collect them all 😂

A lot of trial and error with custom CSS:

.game_frame

{

   background:none;

}

.html_embed_widget button.button.load_iframe_btn {

  display: inline-flex;

  align-items: center;

  justify-content: center;

  gap: 10px;

  width: 300px;

  height: 50px;

  font-size: 0;

}

.html_embed_widget button.button.load_iframe_btn > .svgicon {

  flex: 0 0 auto;

  margin: 0;

}

.html_embed_widget button.button.load_iframe_btn::after {

  content: "Get clicking";

  font-family: "Cracker Winter", sans-serif;

  font-size: 24px;

  line-height: 1;

}