Ask HN: I am overflowing with ideas but never finish anything

Ship incrementally and often

It doesn't have to be "working code". It could be blog articles, conference talks, youtube videos, github gists, colab notebooks, social media posts, etc other places for your ideas to live, be found, discovered.

For example, I've been playing with vector search ideas for a LONG time. I have many unfinished ideas. I can empathize with your feeling. However, you are never really "DONE" - if you shipped something, you'd find the next thing in that area to work on. This loop repeats forever :).

If you scanned, the vector search stuff on my blog, you might not see the "big idea" yet, but I've written incrementally about parts of it, just to get bits and pieces of feedback.

It also helps my focus as it encourages me to keep going in a direction (or not) based on traction.

Journey before destination. Schedule and build in bulk.

Learning Japanese? Take the train/bus/bike and listen to lessons via audiobook, listen to Japanese music, etc

Working on hygiene? Add your habits together in bulk. Do your face care, teeth, etc all together. Position all the products in a line in your bathroom drawer or countertop in the order you expect to use them.

Need to exercise more? Schedule your exercise with something you already have to do (going to the grocery store, walk or bike) or are highly motivated to do (go play DDR, hit the climbing gym, or listen to a video game podcast or 4 while running)

Remember that you are human and have a limited amount of willpower. So don't leave your goals to willpower, make them happen because they have to along the everyday journey.

If you don't have experience in building things, why not just jump on a stabilised project and build something smaller there? Reflect which free software you use the most or find more interesting, and then check how you can contribute.

In the early days of Android apps I saw that there was an LGPL-licensed Java library that could parse Microsoft Access files. I wrote a very simple Android UI for it that one could load a file, choose a table, and then page forward through it. Four days after I had the idea I released the app. At the time there was no other Android app where one could just have an Access database file on your phone or tablet and be able to read it without some convoluted means like net access. So it was popular among its niche crowd immediately.

Not so hard to "finish" - within four days I created an MVP that people found useful. I didn't stop working on it after four days - people asked for the ability to page backwards, which I added. Or the ability to search, which I added. Or for case insensitive search. Or for wildcard search. Or for a recently opened files shortcut if the same file or files were opened frequently. All these things users suggested, all of them I added. I even began sending patches upstream so that the library would work with the then latest version of Microsoft Access.

I had a product which people were using after four days. Because I had a free open source library to use and there was no competition at the time.

David White author of the game "The Battle for Wesnoth" had a similar idea about creating a simple project in a short amount of time ( https://forums.wesnoth.org/viewtopic.php?t=441 ). He had the same problem, so instead of something complex and time-consuming, he started building something simple.

> After a few days or weeks I completely forget about it, having not really made any progress on the project and not having released anything...I intentionally plan to have extremely minimal features, cut corners because this is just a first attempt and yet still it fizzles out.

If you're overflowing with ideas, then work on the simplest of those ideas, which will take the least amount of time to complete. One with even less of the extremely minimal features.

There's nothing wrong with reading about a new language, a new framework, a new API and toying around with it, I do that too as do many people. Lately, when I'm not poking at Android Compose frameworks, I poke at Stable Diffusion or Deep Floyd or LLMs or the like.

But if you want to finish something and never finish something, then a project which takes weeks as you say is out. From your overflowing ideas, pick the one which will take the least time to complete and complete it. If you run into some unexpected very complex tangle, which can happen, then if you want abandon it like the others and try the next simple seeming idea which comes up.