Your graphs roughly marry up with my anecdotal experience. After a while, when you know when and how to utilize LLMs/agents, coding does become more productive. There is a discernible improvement in productivity at the same quality level.
Also I notice it when the LLMs are offline. It feels a bit like when the internet connect fails. You remember the old days of lower productivity.
Of course, there is a lot of junk/silly ways to approach these tools but all tools are just a lever, and need judgement/skill to use them well.
The real game the current administration is playing is to land on Mars before their current term expires. This mirrors the political, prestige, and technology triumph of the Kennedy administration. This is why the BBB Bill refocussed on the Mars mission despite having cuts.
What is being cut is otherwise a symptom of the budget deficit (7%) and the fact that politically they cut areas where there are not republican votes, as politicians obviously try to maintain their voter base as a consideration in their decisions.
Note historically a criticism of the original lunar mission was that USA diverted funds from hospitals and other public programs to fund the mission. So some were bitter despite the triumph.
It goes back to the fundamental conundrum. You have a back of corn. Do you plant the corn, or eat the corn? If AI delivers for America (planting the corn) and USA lands on Mars, these 4k NASA employees will not dwell in the public imagination despite our respect for their commitment, skill and service.
Very insightful - I didn't know about Section 174.
I feel sorry for the overseas labour that a US company startup would be now considering stopping for tax reasons. But I suppose they needed that 15 year amortisation clause (compared to 5 years for a US employee) to stop a flight-to-overseas labour effect of Section 174 (which is presumably 1/3 cheaper).
The framing I see it as is by comparison with the Gold Rush. In those times some made it big (finding Gold) but most of the (reliable) money was made by selling pick-axes.
Here the US is moving from a model where the successful startups cash out resulting in everyone getting paid: e.g. 40% California tax income comes from exits, to being one where the US government takes smaller cuts from medium to failing startups (due to gaining tax revenue of those companies making revenue in their first few years). The drawback is that fewer startups incubate long enough to have a chance of cashing out big.
Note also paying tax as an employee has a better fiscal multiplier to the economy than the owners paying a tax bill. (You buy products and services with your salary more so than wealthy owners on a proportionate basis.)
I find this all quite surprising for the US government, and US innovation culture.
Also I notice it when the LLMs are offline. It feels a bit like when the internet connect fails. You remember the old days of lower productivity.
Of course, there is a lot of junk/silly ways to approach these tools but all tools are just a lever, and need judgement/skill to use them well.