I've lists some ideas below. Some are described in detail, and some aren't. Most are software projects that seem like they'd be useful.
Tools to learn
- Stripe for payments
- AWS lambda for serverless architecture
- WebGL for fast graphics
- Firebase for persistence
- Processing for generative art
- Dask for distributed computation
Short project ideas
- Evolutionary design
- Distribute computation across browsers
- WebGL for fast generative art
- New idea-generation strategies - by prompting with general principles or interesting combinations of existing ideas, artworks, and projects
- Text-to-voice - generate text and read it aloud for various purposes, like idea generation, meditation, visualization/imagination, audio-booking things
- Visualizing music
- Creating beautiful real-world artworks from media like laser-etched glass, wood-burning, or 3D printing
- Turning single images into dynamic videos
- Mental training - memorization, visualization, meditation
- Wordplay - connecting words using various rules, replacing words like mad libs, creating company/band names
Collect, develop, and share ideas.
I get ideas throughout the day, and I'd like to save them. I've tried different tools, but none quite fits what I want. In the end, I want a tool that's really fast to use, allows me to jot down a quick note or develop it in depth, and provides a global view where I can browse all my ideas in seconds.
What I've tried so far:
- Handwritten notes - have to carry around notebook, hard to look at all ideas together, can't edit or move ideas around
- Evernote - can edit and use images, but hard to sort and filter ideas quickly
- Google sheets - can edit and move ideas, see them all together, but iPhone app is slow to load and awkward to edit. Can't associate images with notes.
Here are the features I'd like in an idea-handling system:
- Fast mobile and web access - less than 15 seconds from having idea to recording it.
- Easily extend idea - one app for capturing 5 words of plain text, or 1000 words in Markdown, along with multimedia.
- Quickly browse and compare ideas - can see all ideas at once, in varying levels of detail.
- Quickly add labels - just an extra few seconds to add labels for priority and project.
Here are some other features that might be useful:
- Versioned - can always regain work, or see what changed in each revision
- Analytics - can see when I added ideas, a bit like Github activity monitor
- Open source - Github for ideas. Can see others' ideas, comment on them, fork them, contribute, etc.
- Help generate ideas - maybe incorporate idea combinator
Numbers to words
Convert any number into a memorable story.
There's a memorization technique in The Memory Book where you take a long number, convert each digit to a consonant sound, fill in vowels to make real words, and then imagine a scene making use of those words. The result is that a random sequence of digits becomes a memorable scene that you can decode to recall the original digits.
Here are the rules in the book for turning a digit into a consonant sound.
0 -> C, S (as in celery or salary)
1 -> T
2 -> N
3 -> M
4 -> R
5 -> L
6 -> Ch, Sh, J (as in chew, shoe, and Jew)
7 -> C, K (as in cake or karate)
8 -> F, V, Ph (as in file, vial, or phial)
9 -> P, B (as in pill or bill)
If you started with a number like 44634677129, for instance, you could convert it to the following sequence of consonant sounds - R, R, J, M, R, Sh, K, K, T, N, P. One way to turn that into a string of words would be like this: rare jam rash cook tan paw. There are probably better strings of words to choose, and that's partly why it might be nice to automate this process. In any case, once you have this string of words, you can imagine an exciting story or scene which incorporates them, and which you'll remember.
With a bit of natural language processing, it seems like this process could be automated. Start with numbers, convert them to consonant sounds, and search for words which are consistent with these sounds (they can have any vowels, since vowels aren't associated with a digit). This way, you could just put in any number you like, and quickly get some words back that you can then memorize more easily, and use to recall the digits later on.
Search, annotate, and collect audio more easily.
A listener just presses play. An audio creator does much more - they cut, paste, mix, master, and add effects. I think there's a gap between the two for people who want to listen closely and maybe perform simple edits.
There are a few tasks that should be simple and easy.
- Search - if you have a long recording, you should be able to find the snippets that are important to you.
- Variable playback speed - quickly listen to an entire recording.
- Global audio visualization - see the structure of the entire recording at once. Also, see who's talking and when, or what the music sounds like.
- Retrieval - make it easy to find and replay audio you've searched for.
- Associate listener's notes with audio (at a specific moment, or over some range of time).
- Cut audio and create a new file from just the desired section.
- Bookmark important audio in a single location.
One use case for such a tool is the musician scanning through old material to find things worth reusing. This happens in a jam session, where people record for a long period, with the assumption that most of the recording won't be reused. Finding the parts worth reusing might be time consuming without the features above.
A journalist listening to a debate or interview might like to know who's speaking at each moment. That way, they can quickly navigate a long recording, and maybe transcribe it more easily.
I expect a simple web-based tool could be made to operate seamlessly with audio and video from places like YouTube and Soundcloud, along with files from a user's computer. There are good libraries for handling audio playback and simple edits, so this seems like a worthwhile project.