Really cool! I split my time between cello and software so I'm decently familiar with this line of work. What inspired you to switch careers? I imagine it's not easy to get a foothold as a luthier either. Do you feel like you're taking a big risk?
I switched careers because the software and tech world has changed drastically since I first started, and not in ways that I enjoy or that play to my strengths.
Fundamentally I enjoy being a craftsperson. That is, someone who through training and experience gets really good at something, and then uses that expertise to create new things. Software wasn't always like that, but more often than not I enjoyed it, so I count myself as really lucky. I think the business of software engineering has changed in ways that make it much less amenable to the "craftsperson ethos".
It's not really that hard to start as a luthier - there are 3 violin making schools in the US and you don't really need previous experience, they teach you everything (shout out to The Violin Making School of America - violin making as a newbie can be difficult and frustrating at times but I look forward to school every single day). Since I was a software engineer for 25 years with rather inexpensive tastes generally and no children, and a working spouse, I had saved up enough to not really worry about finances (I certainly needed to downshift some of my expenses, but again, it wasn't that hard for me and in many ways I felt it was liberating - I cook a lot more now, I walk to school instead of rot in traffic, I sold my house which I always hated the maintenance, upkeep and expense of, etc.) I certainly don't have "FU money" but I'm fine being in school for a few years and then making much less than I did as a software engineer.
The big downside that keeps me from using it is that you need to whitelist the contacts you want to be able to call/text, and can’t add new contacts in assistive access mode. Sad, because it otherwise looks perfect as a distraction minimizer.
I’m a musician. For the vast majority of colleagues I know (classical), having to use docker will make it a non-starter. But I’m also curious what your app does, if you don’t mind sharing details!
It's basically a lyric video generator. Right now I'm processing the lyric transcription on a server, but I'd rather users do this locally so I don't have to spend money on that.
I've actually made a few lyric videos with my own music, but I just don't want to have to manage the data of other people.
It’s a good hypothesis. Anecdotally though, I know a bunch of people who have experience playing both, and say that the old instruments can be difficult to play, though sound amazing when played well.
Part of the problem is the potential loss of public value in the form of the undigitized archive, and archive.org provides no solution for that.
Now, is it the responsibility of organization X or person Y to preserve this value? I’m not sure, but that seems to be a separate (albeit also interesting) question.
Hah, I flew planes into ceiling fans too! I also remember scraping my planes against the floor until holes wore into the paper, and seeing how well they could continue flying. There was something really cool about seeing a plane with so much accumulated damage still able to fly.
Is this the type of reasoning that could be used to basic math without errors (without using a dedicated math engine)? I wonder why that type of reasoning seems to struggle.
I think chatgpt is interesting in its ability to highlight the subtle variations in a concept we tend to think of as more of a single attribute (e.g. reasoning) because humans tend to do certain types of reasoning tasks at the same relative proficiency as others. No human can write complex code without being able to add large numbers, so they are often lumped into the same skill category.
I think there just isn't enough mathematical data. English language data is magnitudes more and longer then mathematical data so there is a correlation here.
The more data the better the LLM can formulate a realistic model. If it has less data then the resulting output is more of a statistical guess.
There is an argument that can be made here that the more data the more things chatgpt can copy and regurgitate but given how vast the solution space is I think data at best covers less then 1 percent.
Basically I think that if your data covers say 2 percent of the solution space you can generate a better model then if the data covered 1 percent.
reply