The idea here is to shine a light on these hidden interests and the little (or big!) mental blocks that come with them. If you're already rocking in those specific areas – or you've been there and figured out how to get past similar hurdles – please chime in! Share some helpful resources, dish out general advice, or just give a nudge of encouragement on how to take that intimidating first step.
Let's help each other get unstuck!
1. https://www.reddit.com/r/myog/comments/1k3stln/ultralight_13...
I imagine two languages - Langsam and Schnell - intertwined in some sort of yin-yang fashion. Langsam is slow, dynamic, interpreted, Schnell is fast, static, compiled. Both would be LISPs. Schnell would be implemented as a library in Langsam. If you said (define (add x y) (+ x y)) in Langsam, you would get a Langsam function. If you said (s:define (add (x int) (y int)) (+ x y)) in Langsam, you would get a Langsam function which is a wrapper over a JIT-compiled Schnell function. If you invoke it, the wrapper takes care of the FFI, execution happens at C speed. Most of the complexity typical of a low-level compiled language could be moved into Langsam. I could have sophisticated type systems and C++ template like code generation implemented in a comfortable high level language.
This latter part I managed to partially implement in Clojure and it works (via LLVM), it would be just too much effort to get it completed.
I’ll admit that part of my problem is chronic depression over a decade+. The idea of gamedev excites me, but I have a hard time feeling passionate about anything these days. You definitely need that for games. Hell, I’m barely able to sit down and even enjoy games anymore.
I'd like to go from indoor bouldering to rock climbing, but coordinating with a belayer doesn't seem super interesting and otherwise it's just a matter of expense, gear, and a slight pivot in my leisure time to start going at it.
Otherwise, the skill that seems most out of reach is keeping a job for longer than a year. I'm in a decent spot now, after a year and a half prior of being unemployed, and I feel like this might be my last real shot at a career of any kind. Other people seem to handle it fine, but this is the thing that seems most out of reach. Unlike engineering problems that are made up of abstractions with ways to break them down and piece together systems, keeping a job is as opaque of an abstraction as I'm aware of, that doesn't necessarily depend on a measurable skill or even on anything within one's control. I've never once felt stability or been able to bet on money coming in next year, and if I had the money for a mortgage, I'd be stopped by the knowledge I can't count on an income flow at any time in the future. I'm thankful for what I have and what I've learned nonetheless.
The iPod nano 7g is not designed to last forever... it is hard to repair - the battery can be replaced but the internal 16GB flash memory is a job for professionals.
There are lots of <= 4 inch Android players which could simply run an app, but these are draining a battery so fast after a while it isn't even close to the experience of a specific device.
The main pain point is controlling the device via headphone remote. The iPod supports play/pause, next, prev, vol+, vol- and fast-forward as well as rewind. Nothing comes close to this in android.
However, I was aiming to take one of the newer RISC-V boards with USB-C (like LicheeRV Nano or Luckfox Lyra), add a 2.4 inch display and a battery + gauge to it and 3D print a case, work out a UI via LVGL, but having 2 children is just too time consuming for such a project.
Sticking to my iPod Nano 7g hopefully lasts another 5 years until I can focus on it again :-)
Outside of programming, I'd like to get into welding so I can make some things. I recently learned to use my angle grinder but welding feels like it's out of reach because of not having the right tools and experience.
A job where I can support my family and feel valued/respected. I think that would be cool.
Briefly, one usually formulates the theory of gravity in terms of a a 4d spacetime with curvature but you can also formulate it as a theory of curved 3d shapes if you allow the lagrangian to carry more structure. This is often performed in GR, in fact, by decomposing the metric into a "spatial" and "temporal" part but shape dynamics kind of runs with this idea in an attempt to formulate a totally relational version of the theory of gravity.
Shape Dynamics apparently produces a reasonable theory of gravity which agrees with GR in many situations but forbids, I believe, closed timelike curves, and may be more amenable to quantization since it re-separates space and time.
Anyway, it all seems very beyond me, maybe even if I had the time, which I do not.
I'm a software developer with no real reason to be sewing and lasting my own shoes, but god damn it I'd love to wear my own handmade shoes.
Quantum computer programming. I've dived a couple times into Qiskit from IBM. Also tried to get into dwave and ocean sdk but they never got back to me.
Qiskit tutorials are easy to blow through and i think even understand. But when trying to use it for my own purposes, just never get anywhere.
The other one for me with no success. Training my own specialized predicting AI models. Tensorflow, pytorch, and another.
I certainly prefer pytorch. Super simple to build models on simple stuff.
I'm trying to do something that literally nobody else has ever done. My lack of success has probably a lot more to do with that it's not perhaps actually doable.
Flipside, I might be re-approaching this now that i have the pycharm ai to help me in this progress.
>you've been there and figured out how to get past similar hurdles – please chime in! Share some helpful resources, dish out general advice, or just give a nudge of encouragement on how to take that intimidating first step.
Never be afraid to try. Always dare to fail; you only truly learn when failing. The easier you make it to fail, the quicker you learn.
I have a theory that if society could unlock a hobbyist class of synthetic biology tinkerers it could unleash rapid growth similar to the computing industry. I'm not sure if the problem is technology not being low cost and high safety enough or a lack of social institutions supporting people like this.
I'm imagining physical visualizers that are columns of multiple, discrete light nodes, each able to have variable brightness and color.
The real-time music processing is the hard part (for me) to crack.
There's some standard tricks here: FFTs, bandpass filters, etc.
But I want to do more: Real-time stem separation, time signature and downbeat tracking, etc.
Imagine hearing Sweet Caroline and, when the horns kick in, the whole installation 'focuses' on the horns and bright yellow light jumps between each column on each horn note, before returning to tracking the bass line or something.
I've been noodling on this idea for a long time and slowly digging into the music and CS fundamentals. The rise of LLMs might finally be the piece that enables me to close my intelligence gap and finally build this thing...
There are currently so many cool open source projects in the python ecosystem that involve writing python packages in low-level languages. But unfortunately, I've barely written any low-level code since university, so these projects are effectively out of reach for me at the moment.
However, I do plan on learning Rust sometime later this year and there are number of smaller projects that I plan on working on!
Recently though, I took the plunge and created a Turkish-language book site: https://kitab-evi.com/. It’s all about literature and reading, and now I’m diving into SEO with the goal of getting it to rank in the top search results. I’m learning step by step — experimenting with keyword research, improving on-page structure, and exploring tools like Google Search Console and Ahrefs.
It’s still intimidating sometimes, but I’ve found that the best way to learn is just to start. If anyone has tips or resources for multilingual SEO or ranking content in non-English markets, I’d love your advice!
There's something close called ChoregraphAI[1] on GitHub. It simulates the entire map and does a brute force search for what will get it to the stairs. That doesn't satisfy me because it just skips most of the enemies and takes advantage of things that a human couldn't know, like where a bat will move (bats move randomly, while other enemies are completely deterministic). Also, it can't control the game and is for an old version of it (although those problems themselves seem not that big).
But, I do not know what kind of AI could be written to perform the kind of long term planning and strategizing that a human would do (e.g. isolate each enemy to beat it more easily) and to beat the enemies like a human would (with special, pre-practiced strategies for certain enemies or groups of enemies), and even if I did, I get the sense that this would be really complicated for a first try and that I should gain experience making such an AI for a simpler game, though I don't know what.
It might be a self-constrained challenge. It's just been floating around my head for a while because I don't really know where to start.
maths :/ brain hurt.
i did some digital signal processing in my phd but i need to go through and implement a bunch of things from scratch to learn/relearn and it’ll just be a bit of a grind. i’m avoiding doing that by working on data file parsing / project management utils for the elektron octatrack instead, which is useful, but tangential to what i want to do.
long term would be rad to build software for old synth hardware and the like. sort of like midiquest, but without the price tag.
Creative Writing - Although LLMs seem to be a good help with replacing whatever I am missing. Mostly organizational issues. I enjoy the meat, writing certain scenarios. But fleshing out a whole book I fail from both top down and bottom up methods.
But while I have had a broad interest in compilers for years and years, it feels difficult-to-impossible to actually complete any of it. Part of it's general depression and time management, part is looking at the long story of these projects and the lack of progression: Stuck in a rut of mediocrity.
https://github.com/rkoeninger/ShenSharp https://github.com/jaccarmac/junkcode/tree/79ea647d4ddbe41cf...
Last I checked, there is no fundamental legal barrier preventing this - just an enormous amount of compliance work that has to get done. But as we would be the only real option for 1 or 2 million well-monied people, I imagine we'd be able to start with a very eager customer base provided we could jump through those hoops. And uh, the actual algorithms for index based stock selection are pretty straightforward, too - I'm not Jane Street level clever, so that's an attractive point for me.
If anyone's interested in this hare brained scheme, let me know via the email in my bio. It's been a problem I've been rolling around in my head for a few years now. I have no idea how I would get started, but I'm pretty sure I could do it.
Unfortunately, you can't just sign up for API access to millions of songs. And the streaming apps either don't provide a playback API, or their TOS limits what you can do with it.
Weirdly enough, what helped me break through wasn’t a course or a YouTube tutorial… it was a PDF.
Sounds silly, but I found this open-source project called WFGY, which is basically a semantic reasoning booster for any LLM. You just upload the PDF to ChatGPT (or any model), and it starts interacting in a way that feels like actual understanding — not just parroting surface logic.
No install, no sign-up. Literally a 60-second experiment.
I used it to challenge GPT with questions it usually fumbles (like nested logic, or paradoxes), and it actually got better. The framework even comes with sample questions to test against it. Kind of surreal.
If you’ve ever felt AI prompting is a weird dark art... this made it feel like turning on a light.
I once took a stab at Ladybird browser but had to back out due to the complexity of its build chain. I couldn’t get it run on Xcode/CLion on macOS but would love to give it a try once again.
Does anyone have any tips on getting started again?
I honestly think it could be as important/useful as Von Neumann's CPU architecture, for AI and other highly parallel compute tasks. Every time I start to work on it, something stops me.
For instance, last week I had a good chat with Github Copilot about it, and it was way too nice and encouraging for comfort[2] Unfortunately, you need a Copilot account to read the chat I shared. 8(
I was hoping to get it into TinyTapeout, but they're now only in Europe. 8(
[1] https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...
[2] https://github.com/copilot/share/086351a0-0ac0-80a7-9802-144...
* Bayesian statistics: I know the basics and the theory, but I am not able to understand how to use it in a real world problem
1. Training my own LLM and other kinds of neural networks, e.g. a network for discriminating cheaters in an online game. I feel like it's very hard for an amateur to get their hands on enough training data.
2. Building my own weather prediction engine. Again, accurate prediction requires highly detailed and accurate data to start from and as far as I can tell that's not publicly available.
Also writing my own dbms, but this one I have at least seriously started reading up on and play around with.
Plasma simulation - I really want to understand how fusion and advanced propulsion systems can be simulated and maybe even contribute in some way. E.g. The fusion propulsion being developed by Helicity Space is very interesting. I wish I knew enough to understand their work and maybe do some simulations myself. I tried doing this "Particle In Cell" course I found online but quickly lost interest after a week or so.
I guess I just have a wide variety of interests that come and go. Some of these just seem to be passing interests. I have sort of learned to live with that.
I specialize in computer networking in my day job. Most of what I do is Cisco routers, Cisco switches, and Cisco firewalls. I would be interested in learning more about cellular networks. I haven't put any effort into exploring this for myself. If there is a track similar to CCNA → CCNP → CCIE then it isn't well-known (well, not known to me).
https://www.youtube.com/watch?v=jvPPXbo87ds
and I've been trying to read through _METAFONT: The Program_ and https://pomax.github.io/bezierinfo/ and I keep wondering if I shouldn't just try scripting Inkscape....
I want to do a couple of different things, and not sure if they all fit in one project or no:
- implement a single line font in my current project: https://github.com/WillAdams/gcodepreview
- implement a way to convert arbitrary curves into smooth arcs (for DXFs or G2/G3 arcs for G-code)
- work up an interactive version of METAFONT/METAPOST which allows both programming and drawing
I have welders, metal and woodworking equipment, 3d printers, auto mechanic tools, etc and years of experience and knowledge and interest in building stuff. But I have only a tiny garage to work from. I don’t really have room to do much so it feels like waste.
I’ve looked into renting a small shop but Seattle is so expensive that renting even a one car sized shop space is prohibitively expensive for a hobby (for my income level).
I’m sure I could turn a profit with my work on the side but then it’s a job and you have monthly overhead expenses to cover and all the stresses that come from that.
What is even the actionable step to take? Make more money? That usually comes with less time unfortunately. Moving away to somewhere I can afford a larger workshop at home was the plan until return to office happened for me.
The first is traditional machine learning - not “AI”. I know what I don’t know and I can fake it well enough to talk to a subject matter expert when leading projects and I have a dual math/cs major from three decades ago. But it would take years to be good enough to be at the same level of seniority I am in my existing niche.
The second is more important for my life is learn Spanish well enough to be conversationally fluent. I know some. But my wife and I are going to start living in Costa Rica during the winter and I want to actually learn it to embrace being thier.
E.g. be able to measure distances over a 50cm range to a resolution of ±500nm. Easy for a coordinate measuring machine, very challenging and expensive to DIY.
Validating anything I build means finding a calibration lab willing to check it, which is also rather pricey. I don't have the space, the time, or the money to do this.
Nothing special, prob will be a todo app, but such a steep learning curve to wrap my head around the jargon with everything hosting and docker related.
Wish me luck!
But, there are things I truly have dream but have dedicated almost not effort trying:
* Play (rock) bass
* Draw comic/manga style. I wish to make histories and think I could do it but refuse to write plain words (idiot!).
Basically, art is my unreachable goal
I wish I had the project space, the skills (welding, mechanical), and the tools to build human and electric powered bikes.
I like to see that with known drone sounds, maybe missiles and rockets too.
But that is closed-source.
One viable open-source alternatives is written in Python and uses CNN algorithm.
https://github.com/LonesX/Bird-Song-Recognition
That LoneX lets you add new sounds
My difficulty here is if this problem is interesting or unique, and where to go from here. AI says so (but how much is that worth?)
Fun problems to have! It’s pretty amazing that we can Just Do Things, what a time.
But, that's too much for poor ol' me :). Will just start simple.
I'm also interested in electronics, like building my own drone.
- firearms technology: designing or modifying guns. Feels out of reach because... laws, obviously.
- general ICE engines knowledge: Will probably be obsolete when I get the space and the time to pursue this.
My skill that often feels "out of reach" is building a truly great Go-to-Market machine.
I'm a product-obsessed founder. I can spend all day architecting our 'Workspace OS' or fine-tuning our AI, MAKi. But when it comes to sales funnels, marketing channels, and building a repeatable sales process, my brain just fogs over. I truly feel like I wasn't born with the "sales gene."
After 10 years, multiple pivots, and learning to survive like a cockroach, here's the only method I've found for tackling any skill that feels too big: Make it small.
1. Don't read 20 books; talk to one person. I find one person who is great at it and just ask: "What is the very first, smallest thing you would do?"
2. Don't build a complex system; do one dumb, simple action. I don't try to build the perfect marketing funnel. I write one LinkedIn post. I send one cold email. The goal isn't to succeed, it's just to start and get a single data point.
3. Don't guess what the market wants; ask one customer. I ask, "How did you really find us?" Their answer is always more valuable than a market research report.
4. Don't try to master it in a week; just survive to the next day. The goal is just to learn one thing today, so I can be slightly less clueless tomorrow.
I don't think any of us have the "right brain" for everything. But cockroaches don't have big brains either. They just keep moving, adapting, and refuse to die.
We just need to find the next small, tangible step. That's how we get unstuck.
Hardware: electronics repair, especially vintage ones.
Why I'm not doing it: I'm not sure it would be accepted. I took a lot of lumps for trying to use Hashcash in email, and I'm not sure I want to go through that again, but it does have value. Embedding proof-of-work puzzles in a protocol is a great way to limit abusive requests and patterns. SMTP, HTTP, HTTPS are easy to modify and probably could be done via a proxy. I'm not sure how easy it is to change the SSH protocol, but that would be useful as well.
2) Low-income living space electrification.
I tested this idea out on a friend who works for a Housing Authority, and their eyes lit up. However, they warned me that it would take a few years for everyone to sign off on it.
The original idea was to provide a kit, a bag of parts, that an affordable housing authority could use to improve living quarters and housing for low-income people, and eliminate/reduce the use of gas.
a) Replacing gas stoves with a set of three induction plates. The significant challenges are filling the void created by the original stove, ensuring sufficient power to operate the induction plates, and addressing how to handle the absence of an oven.
b) Filling the hole is easy. This is something a halfway decent carpenter could do, or we could provide an adjustable-size box that fits in such a space, not quite an IKEA flat-pack but roughly similar.
c) Power is a little more difficult. One company is solving this problem by putting in a battery to handle the load. This is possible, but the baseline cost would now be approximately $2,000, just for the parts.
d) window mount heat pump. New York City has funded in-window heat pumps as part of a design project. The problem is they run around $4,000 to $6,000, but an ordinary handyman could install them.
3) Recycling car batteries from crashed vehicles into home power banks.
This project is a bit of a stretch for me. I know people are doing this, but not in the States as far as I can tell. The off-grid solar community has a variety of inverters and solar chargers that may be suitable for this kind of situation, but I don't have enough knowledge.
4) Ad hoc virtual power plants
Many people have rooftop solar. The grid gets overfull on bright sunny days. People who can't have solar often have space for batteries. Work out the instrumentation and accounting so that solar producers can charge batteries, and everybody gets compensated when the grid demands the battery's power.
It seems to me that this would be a great application of distributed system concepts, providing a win for the local community and grid resiliency.
I need a second life to make progress on these ideas