-
bc
An implementation of the POSIX bc calculator with GNU extensions and dc, moved away from GitHub. Finished, but well-maintained.
-
InfluxDB
Power Real-Time Data Analytics at Scale. Get real-time insights from all types of time series data with InfluxDB. Ingest, query, and analyze billions of data points in real-time with unbounded cardinality.
I am a programmer and a blogger, like the author. Unlike the author, I do not monetize my blog in any way (or I assume he does on Substack).
The author is both right and wrong, just about different things.
First, he assumes GPT will get exponentially better. If that were true, all cars would be full self-driving already. GPT will have an S curve or Sigmoid function.
Second, he assumes that GPT will always be able to produce things from differing voices well. However, as more data is used, these bots will only become more homogenous. You can see clues of this when it managed to make a Biggie Smalls rap, but did not with Woody Gunthrie, whose data is probably closer to the mainstream than Biggie's. (This is a gut feeling; could be wrong.)
Third, as things become more of the same bot-like feeling, even from people, those who have their own voice and touch will stand out more. You can see this in his essay, versus the two written by the bot. The voice of the bot ones feels more stilted to me. His feels more natural.
What people need to do is to develop their own voice and touch.
In writing, develop your voice. Write without help. Write a lot. Create a blog. Put random stupid stuff there. Let yourself rant. That random stupid stuff and those rants will not be mainstream, and if it's stupid and ranty, you won't have to care about quality. That lack of care will let you have fun, and that fun will become your voice.
This is how I developed my voice with my blog.
In code, develop your touch. Don't just code; design. Think about concepts. Think about UX and workflows. Iterate until everything is great. Then design the implementation and iterate until everything fits. Then, and only then, code. But as you code, iterate the design to remove things that no longer fit and add things that do. Dogfood your software. Use it for everything you can, even things that don't fit well. Iterate and change some more to make your dogfooding easier.
The end result will be software that is easy-to-use and fits the users' needs well. That fit will be your touch.
This is how I developed my touch with my most well-known project: `bc`. [1]
But you don't even need to stop there. With my `bc`, I spent time supporting users and implementing things for them. Obviously, I still designed changes before implementing them, but I listened to my users and gave them what they wanted within the constraints I had.
The end result is a `bc` that can act like the GNU `bc` or the BSD `bc`, whichever you want. It can also be its own thing with extra features.
Listening to users is another human touch. Make use of it. In fact, that's where I will make a living from my current project: my support for it will be best-in-class [2], and that's something GPT can never replicate because it cannot replicate my brain, my voice, and my touch, even when trained on my writing.
In other words, the author is write that the 90% is useless, but those 90% were the ones where you were at least partially imitating a bot anyway. Those 10% are the human traits.
If you want to have useful skills, be a human, not a fleshy bot.
Footnote: I feel bad for anyone that gets my code or writing as output from GPT. My code fits my head and no one else's. It also uses custom API's that don't exist anywhere else, which means such output would be useless. The same goes for my writing, though to a lesser extent.
[1]: https://github.com/gavinhoward/bc
[2]: https://github.com/gavinhoward/bc/issues/66