Nov. 19, 2021

Sitefox is a back-end web framework for ClojureScript. I built it because I wanted to use ClojureScript to build websites and apps instead of Python and Django. It's inspired by frameworks like Django, Rails, and Flask.

Sitefox Logo

The easiest way to try it is to use one of the npm create scripts.

Use npm create sitefox-nbb for a simple website with no Java tooling dependencies.

Use npm create sitefox-shadow-fullstack for a full stack app using shadow-cljs.


Here are the features it supports so far:

  • Routing
  • Database + Key-value store
  • Sessions
  • Templates (Reagent views)
  • Email
  • Forms
  • Logging
  • Live reloading

There are a ton of examples in the examples folder on GitHub.

If you build something cool with Sitefox let me know!

Nov. 15, 2021


After many days of debugging (and filling out a bajillion forms) Melody Generator is finally available as an iOS app!


It's also available on the Google Play store.

You can still use the web app without installing anything at


Oct. 21, 2021

It's a simple fact that most startup ideas are not going to work out. Just like most tweets never get a retweet, most GitHub projects have no users, and most Hacker News posts never get an upvote. That's the simple truth.

If most ideas are no good the most efficient way to find good ideas is to discard the bad ones as quickly as possible. People talk about "validating" their business ideas by doing market testing, but "validating" has a bias towards a positive outcome. In reality the outcome is usually negative. So don't validate your startup ideas, invalidate them.

I've found it useful to take a scientific attitude. When you have a business hypothesis you need to run experiments to see if your hypothesis matches with reality. In the words of Richard Feynman, "if it disagrees with experiment, it's wrong." Just like Newton poking his eyeballs with knitting needles it helps to be a little detached. Not your retinas though. You'll need those.


This is the story of how I invalidated my greatest ever micro-SaaS idea.

A couple of weeks ago on my morning run, high as a kite on endorphins, I came up with a magnificent idea. I was going to change the world by helping noble open source developers get funded. I'd do this by helping people grow their GitHub sponsors with perks for their sponsors.

I spent the whole day doing some deeper research. There was so much going for this idea. Here is the list of "pros" I wrote down and breathlessly emailed to my entrepreneur friends:

  • viral loop (people link to the URL to use it)
  • proven concept (existing competitors)
  • makes money for people
  • creator economy trend
  • i personally know the customers

Foolproof. I started fantasizing about what I would say in my Indie Hackers Podcast interview when it hits $10k MRR. How one has to stay humble and wait for the big idea. How one has to work hard and stay focused. How one must listen to the customer you idiots, listen to the god damned customer!!!

The idea was perfect. Except for one thing. Nobody actually wants this software. I know this because I invalidated it.

I set a concrete goal. If I can't convince 100 people to sign up for the launch in 2 weeks then I will give up on the idea. John O'Nolan got 30,000 sign ups from one blog post about his idea before building anything. Surely if my idea is any good I can get 100 signups.

Invalidation #1 - friends

The first thing I did was tell some of my developer friends. Their feedback was interested but luke-warm and skeptical. None of them wanted this product. Nobody said "take my money!"

yeah I like your gh idea too but unsure if its good or bad

Ok, that's interesting. Maybe it's just my particular developer friends who don't want this? Friends try hard not to hurt or offend you so "your idea is ok" should really be downgraded to "your idea sucks".

I should also note that one of my friends is an open source developer actively taking donations. They should be the target demographic, but they were not very interested.

Invalidation #2 - search

It's always good to check search traffic. Is anybody out there looking for your solution already? You can check Google search volume, Reddit, and Twitter.

For Google search volume I use Ahref's free Keyword Generator tool and also the SurferSEO plugin.

Ahrefs said there are 80 searches for "github sponsors" per month. Not great. SurferSEO said 880 searches per month. That's better, but I trust Ahrefs more. The keyword is also very broad. Searches for more targeted keywords like "how to get github sponsors" were very low.

A Twitter search for "github sponsors" shows there is a lot of chatter. Out of all of the data I collected this Twitter chatter is probably the strongest pro-validation signal.

A Reddit search likewise shows a bit of traffic for people talking about "github sponsors" but the volume was lower than Twitter.

So I built a landing page

I created a simple signup page where people could get notified of the launch by signing in with GitHub. That should get high quality signups from real GitHub users. It could not be easier. All they have to do is click the "sign in with GitHub" button.


I wrote up the idea in the clearest way I could, explaining the benefits. This also allowed me to proof-of-concept the tech stack and GitHub API integration and make sure I could actually deliver the features. You can see the site at

Then I started the next round of invalidation testing.

Invalidation #3 - landing page

I posted about the site and the idea in a bunch of places.

  1. I wrote a tweet asking if I should pivot from my previous micro-SaaS idea to this.


161 impressions. 4 people clicked through. No signups.

  1. I posted to my local Linux users group mailing list. It has thousands of open source people on it.

1 reply. No signups.

  1. I posted on Indie Hackers with the title "Looking for devs who want to grow their GitHub sponsors".

33 views. 3 upvotes. No signups.

  1. I posted a final tweet as I felt like the first one didn't really communicate it well.


214 impressions. 12 people engaged. No signups.

In the end the MVP landing page had hundreds of open source developers visit and nobody signed up to hear about the launch.

This is a big warning sign that nobody wants this product.

Invalidation #4 - competitors

At the start when I first had the idea I did some competitor analysis. Did anybody else have this idea already and was it working for them? People often see competitors as a bad thing, but usually it just means there is a healthy market already where you can offer a differentiated product.

I re-discovered the story of Caleb Porzio who had grown his own GitHub sponsors. I found his MVP of a similar idea and remembered that I actually signed up for this thing.

It made me wonder why he hadn't posted any progress updates. Why had it not launched yet? I had never received any emails about it. I couldn't help thinking, is it because Caleb didn't get the interest he hoped for?

I also discovered which is a similar idea. Eduardo has 84 sponsors on GitHub, but are they sponsoring him for this or for his work on Vue.js? From what I can tell there are not many people posting on the site.

This research helped me think about differentiators. What would I do differently? These projects both appear to be closed source but I would stay open source. I would also position it differently from Only Sponsors, and offer different features.

In the end this is an invalidation. Both of these people have huge followings on Twitter and GitHub and they have existing sponsorships. Only one of them has shipped and there doesn't seem to be much activity. With my small audience it would be a lot of work on the marketing and distribution side.

Invalidation #5 - target audience

I contacted some open source developers I know who have substantial sponsorship on GitHub. I even contacted devs who explicitly said they have this problem. There was some interest but not huge, and some strong warning signs:

Dev 1:

Currently I don't have any ideas for "sponsor" only content. I tried this for a while with my videos and secret links, but I gave up on that

When I tried something like this with the videos eventually I felt like I was spending too much time on pleasing sponsors with extra stuff rather than my OSS, so that would be one reason not to go there. When I asked my sponsors about this, most of them said: we are sponsoring for your projects, not for the perks, so you don't have to spend extra time on those perks.

As for special content: I did a few "hidden" videos but eventually I also just wanted to share that with everyone. I guess I'm just bad at keeping things a secret or away from people.

Dev 2:

I have done basically nothing to promote it. probably should be doing more on that front but its mostly organic

no rewards no. can't think of anything useful

Dev 3:

I’ve thought this through a few times and talked to a few other maintainers a while ago. My first impression was: The target group (open source devs) love to build things on their own and they don’t have money to spend. That’s both very tough to deal with. Also, I know some teams failed with similar products (probably for that reason).

This is some great feedback. These people are the exact open source maintainer target audience with existing GitHub sponsors. They are telling me no, we don't want this, and other people have failed at it already. Strong invalidation.

Conclusion: invalidated

The strongest message is the signups. Remember at the start I said I was aiming for 100 signups in two weeks? Only four developers signed up to hear about the launch. Three of them are friends and one is my brother.

My hypothesis was "this idea is so good I will get 100 signups in two weeks". Now I have strong empirical evidence to falsify it.

So this isn't the right app for me to build. It could still be a valuable idea for someone. The right person with a high level of passion and commitment might be able to make it work. Maybe Caleb or Eduardo will succeed with it (and I hope they do).

For me though it is nice to have given the idea a good chance and see it through to invalidation. Running this experiment was fun. Now that I have invalidated it I can skip the pain of building all the features only to discover nobody wants it. I feel very good about that!

Sept. 24, 2021

Bootstrapping a one-person software business is a long and winding road. I have the luxury of time and resources on this quest but I know others aren't so lucky. I'm interested in uncovering the secret sauce to launching businesses that work. I love open source and I want to open the sauce, pop!

I have three developments to report. Two experiments and one free tool you might find useful. Here goes.

Experiment #1

First off I want to tell you about an experiment which was a raging success! It was a raging success in the sense that I got an experimental result. The result was I was not able to exclude the null hypothesis. In other words it failed completely.

The Dopeloop Melody Generator online app is now receiving about 9k visitors per month. Most of them come from organic search for terms like "melody generator". My hypothesis was simple: some percentage of those 9k visitors will pay for a native version of the app.

It took me a while but I finally managed to get the paid version of the app in the Google Play store last week. I put a banner on the free online app linking to it and a sales page explaining the new features (mainly "download melodies as a wave file"). Then I went to sleep.

Here's what the banner looks like at the bottom of the app:


The next morning I woke up to the first sale. Success! People were clicking through. If I even got one sale per day then this experiment would be worthwhile. It would prove to me that search-engine-to-product-purchase is a reliable channel.

Alas, there was a bug in this first version, and that sale was refunded soon afterwards. I fixed the bug and uploaded a new build.

One week later and there have been zero more sales. According to the stats very few visitors are clicking through from the online app to the paid app. Why? Are the features of the paid app not compelling enough? Are they on a different platform or OS? Are they just disinterested passers by? Now my task is to iterate on this channel and figure out the answers to some of these questions.

Experiment #2

Ok now for an actual success. Not a raging success but a more hopeful data point.

Last year I released a sci-fi roguelike game called Asterogue. The game took me 1.5 months of part time development to make. I marketed it by tweeting development updates, writing blog posts, and launching on a bunch of different sites and forums.

For my first real commercial game it was a moderate success. It had 38 sales on Android and 27 sales of the Windows version and made about $500 AUD. I was pretty happy with this outcome because I was fully expecting a total flop.

Games are generally bursty. All of the sales are made at the start and they drop off rapidly as the novelty factor wears off, and Asterogue was no different. One thing I kept noticing was the Android version was getting a trickle of ongoing users who were giving it five star reviews.

Super clean and streamlined Sci-fi roguelike without all the cryptic baggage. Definitely a hidden gem that I hope gets more exposure.

Super clean and streamlined Sci-fi roguelike without all the cryptic baggage. Definitely a hidden gem that I hope gets more exposure.

Despite some rough edges people seem to genuinely enjoy the game. This particular review got me thinking. How could I get more exposure for this game?

What I did was put the game on sale for $0 free free free for one week. I have read about other developers doing this and getting a burst of new users. My hypothesis is this: if I can get a burst of new users and exposure, the game will get a second wind of actual sales once the free period ends.

So far I have one day of data and the results have exceeded my expectations. More than 1000 new users on the first day of the sale. None of my games has ever had this many downloads before. Even ones that I gave away for free from the start.

I guess people really like free stuff. Even moreso, they like stuff that used to be paid and is now free. Zero-dollar sales are a great way to drive traffic to your project.

I don't know yet whether this will mean more people buying the game when the sale is over. Will soon find out.

Oh by the way you can still get the game for free if you want to try it:

A free tool for prioritizing

There are always too many things to do. There are more things on your TODO list than you can do in a lifetime. I used to get all tied up trying to figure out what to focus on. Paralyzed by indecision I'd end up not doing anything at all.

The RICE framework is a method of helping you prioritize. The formula is ( Reach x Impact x Confidence ) / Effort. I made a tool you can find at that uses this formula to help you rank your priorities.

You can use it on daily tasks, long term goals, or any other type of time/cost option you're considering. It's completely free and open source. Hope you find it useful, and any feedback is most welcome!

July 21, 2021

We're just over half way through 2021 so I thought I'd post an update about the projects I'm working on right now.

tl;dr: music apps, a new game, and a micro-SaaS app.


A few months back I was feeling frustrated. I had a bunch of projects on the boil and I kept switching between them. I'd switch to whatever was most exciting at any particular moment. I felt like I wasn't making any progress. Context switching is the mind killer.

For a while I tried the 3-month-focus technique from this blog post. I couldn't sustain it. I'm too easily distracted by shiny new projects. I enjoy the thrill of chasing new ideas. Most of them don't pan out but some turn up genuinely interesting results, so I'd like to keep that optionality in my life.

I think I've finally found a balance. Shiny Object Syndrome versus getting things shipped. This strategy has been working well for me this year. What I've done is break down my projects into different areas of interest and then allocate one day to each area. I came up with these major areas of interest:

  • Music tech
  • Game development
  • Open source
  • Decentralization
  • Indie hacking (bootstrapping online businesses).

I assign one day each week where I only work on the one area at the exclusion of the others. I do the most impactful (or interesting) task in each area on the assigned day.

There is overlap between these fields. I am selling my music apps online at the moment so "Music Tech" is also part of "Indie hacking". One of my music apps is open source. Any work I do on decentralized technologies is part of "open source" too, etc.

Here was my side-project schedule for most of this year:

  • Monday = music tech coding.
  • Wednesday = indie hacking.
  • Thursday = open source & decentralization research.
  • Friday = game development.

Here is what my schedule looks like now:

  • Monday = music tech coding.
  • Wednesday = game development.
  • Thursday = indie hacking.
  • Friday = game development.

At the moment there are two days of game dev because I am crunching to finish Smallest Quest. I should be done in a month. Then I can give Friday back to open source software maintainance.

There's also client work of course. I work exclusively on client work on Tuesday. On the other days I do client work too, but I also give time to my side projects.

This schedule gives me enough variety to hold my interest and at the same time I feel like I am progressing each area. Here are the actual projects I am focusing on in each bucket right now.

Monday: PO LoopSync & Melody Generator


I just shipped a big update to my pocket operator Android music making app. I've been working on this app for months. Last year I shipped an SLC and recently I've been pushing updates with new features.

Now that I have shipped that, I will switch to working on Melody Generator. People have been asking for features, and one specific feature I want myself is audio export. The biggest thing to do here is ship native apps. Melody Generator gets around 300 visitors per day. I am hopeful that some of those 300 visitors will convert to paying customers when I release native apps.

I also need to ship iOS ports of both of these apps. I'm working on getting the tech set up to do that.

8 bit music maker


I've been tinkering this fun little music app called 8 bit music maker. It's a tiny tool for making 8 bit chiptune music. I'm not actually work on this on Mondays - I squeeze dev in between other projects. I've been having a lot of fun building this and it is just about at MVP. If you want to follow development of this app I am posting short updates on YouTube.

Wednesday & Friday: Smallest Quest


Last year I released a sci-fi Roguelike called Asterogue. This year I'm building Smallest Quest. It's a new roguelike with the aim of being friendly enough for kids to play. My kids are pretty into turn based games which is what inspired me to make this. I came up with an asset pipeline that lets me hand-draw everything in the game and this has been fruitful and very also very satisfying work. It is nice to be able to just doodle out whatever asset I need in the game. I'm aiming to have this ready by the end of August.

If you want to follow progress on this project I post updates on a Twitter thread here. This is part of a new game development experiment I'm running with 3 friends. More on that later.

Thursday: TweetFeast


My small success with selling apps and getting search engine channels working has convinced me to build another subscription software service. I've built two SaaS products already and both of them flopped. One was more successful than the other, gaining two customers, but neither took off in any meaningful way. There were a couple of things I didn't understand before:

  • Pick a real market with existing demand.
  • Figure out how to reach people in that market.

I think I have answers to those points with my new project. TweetFeast helps you download Twitter data without code or API keys which is a pain point for people who need to do analysis on Twitter data. There are products that do this already but not very well. Lots of people are searching for Twitter sentiment analysis solutions, so I am going to use that as a wedge into the market. TweetFeast provides sentiment analysis of tweets out of the box.

If you want to follow progress on this micro-SaaS project I have a Twitter thread where I post updates about it.

I'm using ClojureScript

Not all of my projects are built with ClojureScript but that's the direction I am moving in. Running ClojureScript "full stack" in the browser and on the backend in Node has been delightful, thanks to shadow-cljs. About half of my projects are now fully ClojureScript.

So that's everything I'm working on at the moment. I would love to hear from you if you have any comments or feedback.

Thanks for reading!

twitter github instagram