You might be working at a place with guilds.

i can't believe this paper didn't cover mycotronics.

also it fails to mention that mushrooms are natures chemists, given sufficient time the can be trained or bio-hacked to breakdown & digest almost anything. the 5th kingdom they are not plants! they emit co2 and consume oxygen.

myco-chitin is a ridiculously interesting in terms of it's tensile strength to weight ratio outperforming virtually all composites including kevlar. the chitin material is a dilectric and can be impregnated with wire strands or other conductive metals to make very interesting unparalleled e-fabrics out of myco-leather.

the "gap" is that mushrooms aren't farmed for aerospace applications, and most of what you find at a farmers market will be either small batch cultivated or more commonly discovered by humans (probably hippies) who spend time wandering through a large forest.

the reality is mushrooms are renewable resources, they can at scale compete effectively with cellulose, and everything you know about useful cellulose -- mushrooms are at least 100x better, but it's easier to cut down a tree than grow a mushroom the size of one.

also the mushrooms require a complex fabrication system and most of the processes we see today are proof of concept, very manual, lacking sufficient automation to compete with plastic & cellulose (so they appeal mostly for novelty, not value).

yes - i've got my own applications in development. no, i won't tell you.

Comments URL: https://news.ycombinator.com/item?id=45935287

Points: 1

# Comments: 1

Issues or discussions related to providing support/coverage/compatibility/workarounds for podman are closed with a terse message. Unusual for an open source project.

Personally I've stopped using openscad recently (*in the last few years). I found it's lack of ability to create attachment/mate connectors significantly inhibited it's ability to compose complex objects. I looked into the amount of effort required to support attachments and ultimately decided that it was the wrong approach and have since migrated to Python & cadquery.

Yes, I do have an expat focused CPA based in the US who ultimately calculates & e-files for me, and however she has a questionaire and still requires me to prepare a summary.

She doesn't correlation transactions to acocunt codes which tend to be very repetitive. In my case many recurring & usage charges on different cards, etc. that need to be categorized, summarized.

As a consultant I've designed and implemented several big data systems which use various ML OCR in workflows. I'm referring to the toil of converting statements from PDF & classifying the receipts.

I'm not especially worried about an LLM hallucinating because I can/could objectively check to make sure the math adds up. If it gets a few categorized incorrectly then it's not the end of the world because mistakes happen -- even with humans. I will still need to audit it at the end.

Even a system which could get 95% correct would potentially reduce my toil from a week+ of emotional pain to a few hours. I wager a model with a fresh context performing a single line item with examples would outperform a human.

This is not something I necessarily need to solve this year .. for now I'm going to have to pay taxes every year until I decide to renounce my US citizenship and that also requires liquidating retirement accounts etc. with severe penalties & horrible tax implications.

I'm not super old, so there is a range let's say 1-50+ years ahead of me (given my age, when I will die) so an effect of automation could be to add a week a year could potentially give me a net result of an entire year. It might also have ancillary benefits of reducing both stress and inflammation, lowering cortisol, etc.

I'm a US citizen who has moved out of the US but still needs to pay/file taxes (despite living abroad for almost a decade, having left immediately after Trump 45 was elected).

I have US investments, mortgages, bank accounts, coinbase, etc. and I find downloading and auditing each of them tedious and a waste of time each year. Like many here I have mild ADHD and when that is coupled with my feelings on the US autocracy it triggers depression. The toil of performing US taxes creates an intense emotional pain -- and like any dev/ops person I want to find a way to automate. I thought this would get easier as the years passed but it's actually gotten progressively worse each year.

Being offshore I experience wildly different targeted advertising and have no friends/social group (beyond HN) to seek advice from.

What (if any) open-source AI tools exist to speed up parsing the data, generation of reports etc. I'd prefer CLI and OSS tools that don't require subscriptions. CLI/TUI preferred.

Examples might be converting PDF's to CSV using OSS, but what else (ex: categorization, expenses, summaries), agentic workflows, etc?

Due to complexities of being offshore I use an accountant who specializes in ex-pat. I don't think software like quicken will be helpful (please correct me if I'm wrong).

Others in similar circumstances what tools do you do to make managing US financial accounts & taxes obligations less tedious/painful?

Comments URL: https://news.ycombinator.com/item?id=42785782

Points: 3

# Comments: 3

I resigned due to the night terrors caused by the cyber security issues I saw everywhere. The more I explored and understood the more sleep I lost.

Needed to find a way to bicubic resize images fast on an embed Nvidia gpu.

Was looking at both llvm ptx & vulkan spir-v as a way to reduce dependency on Nvidia hardware. We ended up using npp with autogen c++ binding.

The rust GPU foundation is a great idea. Lots of interesting possibilities for rust devs.

Rust patterns, because it's advanced type checking (part of memory safety) makes a lot of functional tasks faster for embedded devs. It's also a dream to cross compile and deploy with (cargo(, several times faster and easier than c++.

You need fewer tests and less (often neglected) runtime error handling code because of the contract with the compiler.

I also love how I can write crates for python using pyo3 that data science team can use.

Niron uses synthetic tetrataenite (https://en.m.wikipedia.org/wiki/Tetrataenite) which is a fascinating material. Until a recent 2021 discovery it was believed that you could make it easily, but you'd need to let it cool slowly around 30,000 years! Only a few kg existed on earth from meteors.

But iron nitride Fe3N would also be a similarly good and less exotic choice.

https://hackaday.com/2022/09/01/iron-nitrides-powerful-magne...