Gametje hit the front page of HackerNews!
For those of you that don’t know what HackerNews is, it’s a social news website mainly focused on tech topics with a focus on entrepreneurship and start-ups. It is run by Y Combinator which is a start up incubator in San Francisco, California.
I’ve been an avid reader of HackerNews for 12+ years. It’s part of my daily routine to read and check out all the cool projects people are working on. I’ve largely been just a consumer of the platform, only posting a few replies here or there over the years.
Last week, I posted Gametje on HackerNews and it hit the front page. The “HackerNews” effect and “Hug of Death” are well documented phenomena [1 2 3]. It can direct a large amount of people to your site and even help launch businesses! For the less prepared, it can crash a site leaving eyeballs with only 500 errors.
Luckily, I was prepared for such an event and the frontend and backend survived without breaking a sweat.
Tuesday afternoon - Nov 11th
It was just another Tuesday afternoon, I was taking my son to tennis practice and brought my laptop to do a bit of blogging and work while he was playing. The previous Friday, I had posted Gametje to HackerNews. I was hoping to share it with others before the weekend so they could play some games over the weekend but it didn’t get much interest. I was a bit disappointed but didn’t think too much of it. Back to the present, I had just sat down by the court and booted up VSCode to do a bit of work when I noticed I was getting a lot notifications from Gametje’s Discord server. 20 games created in the last minute? What?!
I quickly went to my PostHog dashboard and saw an onslaught of visitors from HackerNews. I checked my email… It’s on the front page now
- A HackerNews moderator had emailed me, they had re-posted it for me a few minutes earlier and it was already attracting attention.
Terrible timing! There I was, sitting outside, next to a tennis court, wearing my winter jacket with only my laptop and a mobile hotspot to aid me. I was nowhere near my nice setup at home with multiple monitors, keyboard, mouse etc. I watched as more and more games were created. I checked my Grafana instance to make sure the server was holding up (luckily it was). Finally, my son’s practice ended and I rushed home, my phone still buzzing in my pocket from Discord notifications. Time for a fun evening!
Extra Traffic
Gametje normally has 30-40 people visiting it every day. I’ve largely done no advertising so I was fairly content with that especially since it was a part-time gig until recently. Getting attention from HackerNews is a whole other level of traffic.
Here’s a look at my normal traffic:
A few spikes here or there around the weekends. Most people enjoy the “party” games on a Friday or Saturday night. Now here’s that same chart with the last week also included (using a slightly different scale):
You’ll notice my “normal” traffic is nearly a flat line compared to the traffic I got last week. Crazy to see!
It was an exciting evening seeing all the new games being created and people experiencing my hard work from the last 2.5 years. Most games were played with only one human player and multiple AI players. It seemed to be people checking it out on their own without inviting others to join. This was expected and also part of the reason I made an effort to support AI players in all games. I hoped people would return to try it with their friends and family later as it is much more fun with real people. Throughout the evening and night, around 1200 games were created (700+ with actual play time). At one point, the Discord webhook was timing out with the number of requests it received.
Considering HackerNews is primarily a US-based site, I was also quite surprised of the global reach. I had a post about a month ago about how Gametje had gone global and had reached ~69 countries since I made it publicly available about 2 years ago. In the last week, it’s been accessed in ~103 countries! Here’s a map:
It was amazing to follow all the different visits coming in. It got almost all of North and South America and a good chunk of Europe and Asia!
Server Performance
Gametje’s frontend is written in React/Typescript and hosted on Cloudflare. This amount of traffic is a drop in the bucket for a CDN like them so there were absolutely no issues there with staying up or slower load times. It reached around 1.5 million requests on November 11th alone, most occurring after 16:00 when Gametje hit the front page.
Gametje’s backend also handled the extra traffic without any issues. The CPU maxed out around 4.5% (0.36 cores out of 8). I also got very few errors in the logs. Everything seemed to work well. Here’s a graph of the CPU usage:
If anything, it seems I am over-provisioned. I had previously performance tested the backend using Gatling and got up past 300 logins/second so I didn’t really expect it would fail. The only thing I had not properly tested was the WebSocket layer which is used for interactivity with the games. Luckily it seemed to function well under load. Overall I was very satisfied with how the servers performed.
Feedback
You can check out the HackerNews post here. The summary of the feedback had a few key takeaways:
- People asking about the name (which is not unexpected).
- People complaining about the UI (which is fair, I’m primarily a backend developer).
- People thought it was a good idea and under-served market but the execution needed a little work.
The name
I’ve been toying with renaming it for a while. On one hand, I like the name ‘Gametje’ as it tells a bit of the story behind the project itself. It’s also quite unique and makes SEO very easy. If you Google it (do people still do that?), it is all of the top hits. I think including the word “Game” is also crucial to its success. The feedback I got from HackerNews and from others since I created the project was that Gametje was not easily marketable internationally. It doesn’t really roll off the tongue and the tje is an unusual sequence of letters outside the Dutch language. I actually purchased another domain name recently to possibly transition to: gamebuffet.eu. GameBuffet is easier to remember and tells a bit more about the site. It’s meant to be an “all you can play” platform with many games for one low price. I also heavily use food related avatars so it seemed fitting.
Unfortunately the .com and .net were not available but since I am based in the EU, it was the next best option. I’m still evaluating if I want to change the name given I already have a decent presence with the name Gametje now. For now I am going forward with Gametje.
The UI
Honestly, I am pretty happy with where the UI is for all of the marketing pages. I worked with an UX designer earlier this year to give the site a face-lift. The games themselves are still fairly bare bones in the UI execution and the flows around hosting could use some work. All of the games were created when I was still working on Gametje part-time. I mostly had focused on getting the games playable rather than looking amazingly polished. In the last couple months, I’ve been trying to make the design more playful and fun but it still has a long way to go. If you are a designer reading this and interested in helping, please reach out!
Traffic Ripples
Given this blog post is a week later, I was able to see a some sustained traffic after the initial onslaught on Tuesday. Around ~500 Games have been created excluding the HackerNews ~24hrs. Once the weekend (November 14-15) rolled around, I saw that there were many more games being played with human players. I had a few groups of 10+ people playing! It was nice to see people returning after the initial interest. Canvas Clash and Imaginarian remained the most played games. They are also the first two you see when visiting the site. Currently traffic is still around 5x what it was prior to the big day. We’ll see if people continue to return going forward.
What’s next?
Given that I have quite a bit more traffic now than before. I need to continue working on the guest account conversion. Right now you can play for free for a few days but there isn’t a compelling reason to sign up. I wanted the initial barrier for entry to be low to get as much feedback as possible. Now that users can see the benefit and fun of the platform, I need to figure out the best way to convert them to regular users.
I could restrict the types of games a guest can play or maybe the amount of games in a day/week. Maybe only offering the newest games to regular users? Still thinking of other options. Regardless, I need to work on an overlay to show to the hosting user that helps them create an account seamlessly. I would still like continue offering a free option as it gives new users an easy path to try it out without any hassle.
Let see what we can do with this momentum. If you made it this far, Thanks for reading!