Ask HN: Show your failed projects and share a lesson you learned

I spent a year during and after grad school writing a blog and building a tiny community on healthy living for elderly people. I remember a particular article I wrote on eating pineapples for healthy bowel movements really offended some in the community.

My grad research was on AI and robotics for elderly. I thought I knew my audience. How wrong I was. I didn't even reach out to talk to my audience, for crying out loud.

10 years later when I was working on a project for smallholder farmers in Mozambique. I made the point of moving myself to Africa and working alongside them for weeks. https://www.quantisan.com/facebook-loving-farmers-of-mozambi...

I didn’t finish my PhD, what’s wrong with it? I don’t understand how somebody can judge that you didn’t finish a complex project because your priorities hanged at some point. PhD is a job and people quit their jobs sometimes, its ok.

Had the opposite idea a while back. Basically you tell people it's a single player Mario-like game but every enemy you meet is another player (flipped along x-axis so both walks to the right and others looking like enemies).

In Super Mario you don't actually need to kill anyone to finish the first level so I thought it would be quite fun to see what happens.

You might have to prime them with that the AI-system is broken and that is why enemies have so different patterns... or maybe people play very similarly?

The version of this phrase that I am familiar with is “perfect is the enemy of done.”

What is the fundamental issue why you keep switching tech stacks?

I'm not an architect or senior developer, and not saying my take is better, but my considerations would be very different.

First, I would want to have a reasonably clear idea of what you want. My first idea when I hear 'stackoverflow for flash cards' is that you want collections of flash cards. I suppose you need some discovery mechanism, and ways for users to upload and share collections. You mention syncing, I suppose that means that every user always has the last version of a collection. What happens if a user deletes all or most of the flash cards in a collection? Will it be lost for the user or will (s)he be able to use an older snapshot of the collection. This would probably make things more complicated.

Then, I would consider the architecture. It is natural to go for a client-server architecture (and at this point I would draw two circles with two arrows pointing both ways between them). This simple architecture is enough for most applications.

The arrows in this diagram represent dataflow. The choices you make determine the actual data that is sent "over" the arrows. In this case you would need a strategy to distribute the flash cards and update them. So, you define the types of messages that the client and server need to handle and define some data structures for them.

So, the next step is to pick an implementation. An obvious choice for the arrows is HTTP. The client and server can then be implement in any programming language, not necessarily the same. The server can use some DB in the backend. So, again you need have some dataflow between db and server, so you can refine the diagram a bit, and define the data that goes over the arrows.

So slowly you can build up a detailed architecture. Note that it is mostly tech-agnostic, implementation should be possible in most programming languages and you can even make multiple types of client.

Even if you decide to redesign, with this method it is easier to keep your design clean and salvage some parts of protocols/structured that worked well enough.