Back Propagation

When training some types of artificial intelligence software — errors in the output of the software are sent back to the program so that it can hopefully correct itself or minimize the errors to acceptable levels.

This is post is about me doing the same on the software that runs in my mind — the one that evaluates new ideas for the sake of investing in them, working on them, promoting them or just for the fun of it. I have said about many ideas / products “this thing will never work” — only for it to — I want to send back some feedback so that I change.

I have have reviewed approximately four different ideas in the last 2–3 years that at the time I thought were non starters. Last week I revisited them and saw they were actually interesting — if executed well could be rewarding.

Ironically one of the ideas is Medium — the very medium on which I am writing this post. I thought the world did not need another blogging platform — we had wordpress, typepad e.t.c Which can be collapsed to I did not think anything better could exist — until I wrote on Medium — and now I recommend everyone to do so — it is beautiful and elegant — it encourages you to write — no annoying sidebars, plugins to manage, comments spam, no fonts to think about — it is simple.

I got convinced about Medium when I used it — and later when I read the mission of Medium. I was hooked and I still am. So the thing to send back to my software is simply — being curious to try something out and suspending the reasons why it would not work — increase the curiosity appetite.

The other three or so ideas I dismissed them because I thought they wouldn't get any customers — because I imagined myself using the product and I could not see myself or my peers using it. Turns out I wasn’t the target market (even though the product promoters where selling to me) — so naturally I concluded it won’t work.

Those ideas haven’t worked yet — but I now think they can — what has changed? I’ve learned more about customers needs and situations — and seen how they can work. My fault here was that I thought I was the customer, which is a specific case of the more general case — I did not understand their customers well enough — or their problem. The pattern change I’d like to make is to listen to customers more, explore their situation and their problem.

These two lessons seem deceptively trivial — learn about customers and be more curious but let us see if they improve my software.