there are great empire builders but they are few and far between
To paraphrase Einstein
Instead of me trying to emulate the empire builders, I should just be myself and that means finding / scraping / earning a ton of autonomy to build products and systems I want to build
Thanks for your great newsletter. I would love to hear your opinion about the failure of Kite as a business.
I agree with the Kite founder that the code technology was way too early to be built as a business in the mid 2010s. It was at least 10 years behind prime time. To give a perspective, 2014 was the time when the first academic paper introducing the attention mechanism behind all used Transformer models was released (https://arxiv.org/abs/1409.0473).
But I don't fully buy the founder's argument that "making their developers 18% faster when writing code did not resonate strongly enough" with engineering managers. I am a happy paying user of current code writing assistants like Copilot, and I use them for both job and personal projects.
The time savings with these tools are massive (even though they make blunt mistakes sometimes), and I am surprised that engineering managers would not want their developers to be more productive when writing code.
I'm curious to hear your thoughts as a former engineering manager.
A timely reminder for me
there are great empire builders but they are few and far between
To paraphrase Einstein
Instead of me trying to emulate the empire builders, I should just be myself and that means finding / scraping / earning a ton of autonomy to build products and systems I want to build
No more mindless mimetic desire copying
Hi Louie,
Thanks for your great newsletter. I would love to hear your opinion about the failure of Kite as a business.
I agree with the Kite founder that the code technology was way too early to be built as a business in the mid 2010s. It was at least 10 years behind prime time. To give a perspective, 2014 was the time when the first academic paper introducing the attention mechanism behind all used Transformer models was released (https://arxiv.org/abs/1409.0473).
But I don't fully buy the founder's argument that "making their developers 18% faster when writing code did not resonate strongly enough" with engineering managers. I am a happy paying user of current code writing assistants like Copilot, and I use them for both job and personal projects.
The time savings with these tools are massive (even though they make blunt mistakes sometimes), and I am surprised that engineering managers would not want their developers to be more productive when writing code.
I'm curious to hear your thoughts as a former engineering manager.
Good read as always.