It's not directly comparable. The first time writing the code is always the hardest because you might have to figure out the requirements along the way. When you have the initial system running for a while, doing a second one is easier because all the requirements kinks are figured out.

By the way, why does your co-founder have to do the rewrite at all?

I find the opposite to be true. Once you know the problem you’re trying to solve (which admittedly can be the biggest lift), writing the fist cut of the code is fun, and you can design the system and set precedent however you want. Once it’s in the wild, you have to work within the consequences of your initial decisions, including bad ones.

... And the undocumented code spaghetti that might come with a codebase that was touch by numerous hands.

You can compare it - just factor that in. And compare writing it with AI vs. writing it without AI.

We have no clue the scope of the rewrite but for anything non-trivial, 2 weeks just isn't going to be possible without AI. To the point of you probably not doing it at all.

I have no idea why they are rewriting the code. That's another matter.

This is interesting, how do you get it done? From what I know CAD tools generally don't support text file, only binary blob which is LLM unfriendly?

Do you consider adding support for AutoCAD or AutoCAD vertically integrated software like Civil 3D?

The conversation itself is sent to the LLM in regular text, and in addition it sees the feature tree (also text) and often a screenshot of whatever the current model looks like. This is usually enough for the model to know what's going on.

Yes - we're likely looking into other 3D systems in the future.

When I started my company, we suddenly found that we were in a good small fortune, not enough to be millionaires or billionaires, but enough to get people to run the business semi automatically with very minimum input from the founders.

I took a semi retirement approach to the business, there really wasn't a lot of things to do, my role was sort of just "managing" programmers. I got so much free time that I could even start a second business on the side.

Despite my best ability to stretch my work, I couldn't even fill up half of my working hours. One would have thought that this is heaven. But the time I was most free was also the time I was most miserable. I wasn't happy, I was gaining weight, I was perennially asking myself why the business couldn't be bigger and I couldn't sell it, so that I can be real millionaires and billionaires with financial freedom!

Then fate intervened, the sudden fortune disappeared and I no longer had the luxury of just "managing people"; I have to do hands-on. And it was this activity, the feeling that I was contributing to something, that I was writing code again and actually building stuffs, that made me happy again.

Today we are bigger than what we once we were, but still, I am writing code and pretty much hands-on.I vow that I will never retire, even though if I could. Because it's the meaningful work that sustains life and provides happiness. Being able to work on it is a luxury that I will never want to give up, ever.

"Because it's the meaningful work that sustains life and provides happiness."

For you.

For me, it is having the time to do what I wish. Currently helping a friend with recovery after a major surgery. Next month, who knows?

No, it's not at all the same as "meaningful work".

At least in part, I do not need the attaboys or regular 'sense of accomplishment' that one get from plate-spinning or other meaningful work.

You agree more than you disagree. Meaningful might be something else for you. But I bet deep down it is doing something for others (clients, coworkers, friends, family, even strangers) that gives meaning.

This is an astonishing revisionist take on the reality on the ground.

Israel unilaterally disengaged from GAZA in 2005 and pulling out generations of Jewish settlement in the process. By 2006 GAZA has zero Jews, and 2007 Gazans elected HAMAS who fired rockets at Israel because they want to free Palestine from the river to the sea, AKA eliminate Israel. October 7 attack is a culmination of that, and between then and now, HAMAS didn't forget to build their military base in the mix of civilians and using civilian targets as shield. So that they can blame Israel for every single Palestinians death, including the death cause by their own firing.

The situation in west Bank is qualitatively the same.

No, protecting your people from terrorist is not apartheid, and Israel has no interest to build iron beam and/or build wall--which the west misinterprete as apartheid-- if the neighbors had no intention to eliminate them.

There are 2 million muslims living in Israel.

There are zero Jews in Gaza -- not even just living ones, they had to remove the long-buried dead ones too.

The millions being starved purposely are for all practical purposes (and by the territorial recognition of several countries if that means anything...) also living in Israel. More to the point, I am not sure what that means for the discussion: I am sure I'd have evacuated my own citizens too before doing this.

The least important point is technical, which is that I said Palestine lost many homes to Israel over the years, not that every (just many) Palestinian lost their home. For what it's worth to the symmetry argument there were Jews happily living in Palestine before 1900, but let's not get distracted by the question of how good my opinion is when there's an obvious wrong anyone can agree with staring straight at everybody.

> For what it's worth to the symmetry argument there were Jews happily living in Palestine before 1900,

Someone's never heard of jizya, or the repeated slaughters of Jewish civilians in the century prior to Israel's creation: https://www.fondapol.org/en/study/pogroms-in-palestine-befor...

Not to mention what some backward, cruel warlord did to much of the Middle Eastern population starting in 622:

"Alongside his campaign against the Quraysh, Muhammad led campaigns against several other tribes of Arabia, most notably the three Arabian Jewish tribes of Medina and the Jewish fortress at Khaybar."

https://en.wikipedia.org/wiki/Military_career_of_Muhammad

And it never ended. It's continuing to this day.

> Israel unilaterally disengaged from GAZA in 2005

There were over a thousand gazans held without charges by the israeli military on oct 7th. That is not what disengagement looks like.

Israel has military bases in cities.

> No, protecting your people from terrorist is not apartheid

I’m quite sure a white south african could have said this same sentence pre-1994.

Thanks for the comment that exposes the nuances. As a non physicist who is interested in physics, I would say that the predicament is that we simply don't have enough data to tell how to improve our physics theory, but we have too many physicist working in the field, so ideas just get staled.

Is that a correct assessment?

Not exactly, I'd put it as follows: 1. In phenomenology (the science that looks into extending Standard Model) we'd definitely benefit from more data (with most viable data sources, in my opinion, being not the colliders but the detectors of extraterrestrial particles such as PandaX or Fermi-LAT). However, this doesn't stop us from getting new results, narrowing the window of possibilities for the properties of the Standard Model extensions. This "sorting a haystack in search of a needle" approach may not sound too cool, but this is exactly how we discovered the Higgs boson, for which we looked for more than twenty years. 2. In hep-th we currently lack either enough data (which we can't really collect now), or an ideological breakthrough in solving one of the hard problems. Unfortunately, the problems are indeed very hard, so that one of them (Yang-Mills existence) is a Millenium Prize problem. This caused the de-facto shift of hep-th into applied math, so now hep-th is not constrained by the data problem, and is developing as actively as an area of applied math can.

But software engineering problems are more fuzzy and less amendable to mathematical analysis, so exactly how can those AI policies developed for math be applied to software engineering problems?

Not sure which way the difference puts the pressure. Does the fuzziness require more prudent policies, or allow us to get away with less?

Don't use them for the parts that are fuzzy.

I mean it should be obvious that making executive decisions about what the code should do exactly should only be left to a RNG powered model if the choices made are unimportant.

A Civil 3D plugin (Genabler) that will include all the network catalogs and collate the Civil 3D styles for civil engineers to use. There are some out-of-the-box catalogs and styles shipped with the default installation, but they are quite limited and fairly well hidden—which is not surprising, given that Civil 3D is a huge beast. As a result, they are not commonly used. When people think about Civil 3D, they often assume it requires BIM modelers (in a sense, just glorified drafters) to create all the necessary catalogs and styles, and to assist with their use.

My Civil 3D plugin will:

1. Make standard, market-compliant catalogs and polished styles available to engineers at large. Think of it as the WordPress theme provider equivalent.

2. Make the entire process easy and painless through the plugin, with prominent buttons for quick access.

If the plugin is done well, there will be less need for BIM modelers, since for a fee, engineers could simply purchase catalogs and styles that are so easy to use they require no technical training.

As a side benefit, I also get to explore how LLMs can help me write code. It has been a while since I last updated my AI usage policy [0], and I look forward to revisiting it.

[0]: https://civilwhiz.com/my-ai-usage-policy/

I think the fact that DeepSeek trains on competitor queries (i.e., distillation) — along with using banned Nvidia chips — helps explain how it can achieve such low training costs (USD 6 million vs. billions) while delivering only slightly worse performance than its American counterparts. It also undermines the narrative that DeepSeek or China is posing a serious challenge to the U.S. lead in AI. The gap may be closing, but the initial reactions now seem knee-jerk.

That the discussion has being hijacked and shifted to moral superiority is really unfortunate, because that was never the point in the first place.

These models never cost billions to train and I doubt the final training run for models like GPT-4 cost more than 8 figures. 6 million is definitely cheaper and I would attribute that to distillation.