Large companies are incredibly unproductive and inefficient.
That said the unevenness of contribution isn't strictly a large company phenomena. Small companies have the same uneven distribution. I've worked at two startups with about 4 people total and people were not equally productive.
That said, this is not necessarily the goal and productivity is also very hard to measure. It's doubly hard to measure across different types of work. One person can code up a greenfield back-end for something in 3 days while another can spend a week fixing some elusive infrastructure problem.
Not everyone is as good at everything. So we do have engineers who truly are much better than average. And in large companies most are average. But that is just one factor here.
1:1's can add value or they can not add value. Large companies can't just be flat so someone needs to manage people. A good manager adds value, a bad manager might subtract value, but that's orthogonal.
Is it demoralizing to work for a big and inefficient company? Sure. Is it more demoralizing for people who are motivated to get things done and are good at it? I think so. Go start your own company?
36 hours per week. 25 days vacation (going to 28). Pension contributions. You can buy extra leave. Epic location, fun job, decent salary for the UK (where e.g. you don't pay for healthcare)...
Talked to a German guy who was here on holiday recently. When I told him that in the US it's typical to get two weeks vacation when starting a new job, you should have seen his eyes bug out. It was hilarious.
Karate traditionally has 3 components, Kihon, Kata, and Kumite ... I would try going to some more traditional Karate Dojo and asking if you can practice with them without sparring. Most of the class isn't sparring anyways and if the others spar then you can just practice your Kata or Kihon while they do that. While sparring is important it's possibly the least important component in many traditional schools.
Whether or not you'll be able to progress in belts/ranks is different but who cares. Wearing a white belt forever is probably better for character development. If your goal is the belt just go to the store and buy a belt.
There's also archery and I'm sure there are many other options for no-contact martial arts training. If fencing works for you then there's also Kendo. There's Iai-do which is just about drawing the sword. Lots of options in theory.
Yet another option is to wear some sort of body armor during sparring. There are suits you can wear that protect you from pretty much any human form of attack. We use them for some self defense training.
In BC, Canada, the recommendation is to take cover under a desk e.g.
My personal take is that if I have a chance to make it quickly out of a building I prefer to be out of it rather than having it collapse on me. It's one of those things where the recommendation probably is the right thing for the general population and all expected earthquakes but I'm optimizing for something else. You have 10-20 seconds realistically. But yes things falling off the walls and off buildings and anything that can fall on you is a risk. In my office or home e.g. I'm typically on the ground floor within seconds of an exit so my mental preparation is to avoid the building falling on me. Maybe it's the wrong calculus but hey...
What is that protest going to get us? We'll convince or force business leaders to not use a cheaper/better tool and protect our jobs? And nobody else in the world is going to pivot either? And our companies will remain competitive?
Software engineers have always adapted to new technologies. New languages, frameworks, native apps, browser apps etc. So far this doesn't seem to be close to completely removing us from the loop.
If you are smart, educated, and can adapt, you'll figure it out. The economy has to find some stable equilibrium and it's not a zero sum game. Everyone in the economy getting a paycheck is also a consumer. With no consumers there is no business. The companies who are using AI and become more productive can do more things that before were not profitable but now are. Some of the people who are getting laid off are going to start new businesses and hire people. These things always cycle, and they basically have to.
Lines of code has always been a terrible metric. But all else being equal it is a measure. If all else is not equal, which is usually the case, then it's not.
A lot of the focus has been on AI recently.
Three years ago we didn't have software where a non-software engineer can describe what they want in English and get working (-ish) software generated by other software? Is that not "software has gotten a lot better"?
Other than that I'm not sure how we measure "software has gotten better". New applications? More features? How do we measure sloppier? Is Google Maps suddenly taking you the wrong way more often? I'm not really doubting your subjective experience but seriously how do tell? I mean a doc is a doc and a spreadsheet is a spreadsheet.
We're also only about 10 months into models that are powerful enough to potentially make a bigger difference and we are still figuring out how to use them best.
I personally don't view coding agents making software as "software gotten better" you are comparing a tool and the end result, these are two different things. Agent you use going down and your product going down mean two different things to you customers. I will not deny that we made incredible progress in coding and hell, even design over the past 3.5 years, this technology is here to stay.
That being said while I agree that measuring better quality of software is vague (part of the reason it is hard for models as well), there are universal things I believe every engineer will agree on. Reliability, uptime, customer feedback, legibility of your engineering, performance, these are things we often optimized for. Google Maps is a bit of a strawman because neither of us (unless you work on it), knows how much agent code there is, I think it is likely that it's little since it was working fine prior to 2023. I could bring up github reliability as an example, given how much copilot usage they promote at MS, but once again only folks there know for certain. I do, however, see scores of various AI powered SAAS that looks like it is in a perpetual MVP state. I think you are right in that even if agents give us "good enough" results and we can swallow failure rates and our increasingly lesser understanding of what we, or more so model, created, then it is still progress overall, but this is progress not to human-AI collaboration but to AI-only engineering IMO, this is good or bad depending on how you view the future.
I'm a scientist and most of code I currently write is somewhere on the intersection of critical software and machine learning, squaring these two is not easy and I guess the way I was taught to reason about engineering informs my opinions on this. Maybe it's just a matter of time before codex can help here in an unconstrained manner as well, but I am skeptical at the moment.
My point was "look at what computers(software) can do today vs. 3 years ago" - for everyone. You are saying that software that the arborist can have ChatGPT write to help it draw the garden isn't the same quality as a team of software engineers would write manually (I think). GitHub is (mostly?) software for software developers. Most software in the world is software for random people. Nobody(tm) cares about the quality of GitHub. The "has software (computers) gotten better" has to be measured from the perspective of the consumer, not the perspective of the software engineer, and nobody using AI is going to tell you "computers now are worse/can do less than they were 3 years ago". At least that's my thesis.
If AI today can make you more productive that's already progress. If it can't then maybe it makes other people more productive.
> Lines of code has always been a terrible metric. But all else being equal it is a measure.
A terrible metric is _worse_ than no metric. A terrible metric can _only_ lead you in the wrong direction. "No metric" means saying we don't know, and that leads us to stop and reconsider. But we've taken "move fast and break things" as a mantra, and we'd rather run towards any direction than stay still.
Using LoC as a metric for quality of LLMs will promote LLMs that write more code. It's better to say we have no way to compare different LLMs than it is to say "let's use the LLMs that produced more LoC because at least we can measure that". We, as an industry, should be focusing on developing better metrics for quality, not on improving LLMs based on known-bad metrics. We should be turning to the computer scientists, not to the venture capitalists.
When a pundit talks about how many lines of code an LLM has created, we should lose all respect for them. It's as if someone talking about physics measured the phlogiston, or as if a doctor started measuring our skulls. We know these theories don't work, and anyone using them should be mocked.
I also can't help but notice they didn't mention how many tokens were burned, or how much that translates to in terms of cost over the 5 months at enterprise AI prices. I'm going to guess this wasn't a cheap demo.
> Is Google Maps suddenly taking you the wrong way more often?
Funny you mention that because I had that issue in a cab just yesterday. Google decided to drive us of the main road to a series of small roads which happened to be a dead end. My guess is that the AI decided that this is a shorter road? less busier road?
That being said, Google maps have been gradually degrading. Most notably, its search function is quasi-broken now.
You must be joking re: peaceful before US and Europe. The first crusade was in 1099 for those who don't know the details. We had the Byzantine-Arab wars, Fatimid civil wars, Turkish invasions... Ofcourse we had the whole spread of Islam "by sword". Don't forget it was the Roman invasion of the region in 63 BCE that resulted in the mass murder and expulsion of Jewish people from Israel after the Bar Kokhba Revolt...
Are you talking about the Ottoman Empire? Pretty violent.
Anyways, I can't cover the history of the region in an HN comment...
So you're arguing the crusaders brought peace to the middle east?
This history is so vast I can't even begin to think about how to compare. But one thing that feels odd to me is how people think of the middle east as somehow separate/far from Europe when in fact it's basically the same neighborhood. The Greek and the Romans were there. Under the Ottoman Empire, Muslims from present day Bosnia moved to present day Israel: https://en.wikipedia.org/wiki/Bushnak
Don't forget that Christianity came from the middle east and ofcourse Islam.
The Ottoman Empire ruled vast swaths of present day Europe. Spain was under Muslim rule until 1492.
It's all one big mesh. Just yesterday I learnt that many present day Yemeni trace their roots to the Levant. Very different than farther regions like Afria, China, India and ofcourse the Americas, Australia etc.
You might be right on the regime change being fantasy but those things are not predictable and we don't know the details.
Where you're definitely wrong is on the "diplomatic agreement that worked". Iran continued to enrich violating the agreement, the agreement was time bound and not indefinite (and would have already expired anyways), and it enabled them to sell oil and raise a lot of money to fuel their wars, missile programs, nuclear programs and other ambitions.
> Where you're definitely wrong is on the "diplomatic agreement that worked". Iran continued to enrich violating the agreement...
No, actually it is you who is wrong. Iran absolutely complied with the JCPOA. It is after US withdrew from the agreement that they pursued enrichment further.
Yup. "The U.S. certified in April 2017 and in July 2017 that Iran was complying with the deal. On 13 October 2017, President Trump announced that he would not make the certification required under the Iran Nuclear Agreement Review Act, accusing Iran of violating the spirit of the deal..." [1].
Wrong. That is not the question. It came to light later that Iran hid sites, activities and materials. The 2017 certification is not relevant. They were still violating it either in letter or in spirit, they had no intent of stopping the pursuit of nuclear weapons, and at the most charitable interpretation (and no way the Iranian regime deserves that) the agreement was time bound and would have expired already.
Why was Iran under sanctions in the first place? Sponsor of terrorism. Oppression of its own people. Messing with Yemen, Syria, Lebanon (and the list goes on). Only in Syria they helped Assad murder 100's of thousands of Syrians. The Yemen civil war. The murder and abuse of their own citizens.
Iran had an easy way of not getting sanctioned. We didn't need the JCPOA. What we needed is Iran to cease the activities for which it was getting sanctioned.
We had a "diplomatic solution for Iran" is total nonsense. Obama messed this up just like he totally messed up the entire middle east. Iran trained and supplied Hamas which led to Oct 7th. Iran trained and supplied Hezbollah. Iran developed and built their ballistic missile program to attack all their neighbors. With what money/resources? With the money Obama gave them in for cheating on this agreement. If you have western interests in mind than the Iranians are laughing at you for being a fool.
> No Active Bomb-Making (2016–2019): Neither the U.S. intelligence community nor the IAEA found evidence that Iran was spinning secret centrifuges or actively manufacturing a weapon at these sites while the JCPOA was in effect. The traces found were leftover from the pre-2003 weapons program.
Thanks. You proved my point. Did you even read the first article you posted?
> "...the material in question is probably from a clandestine project that was first discovered in 2005 and reported by the IAEA the next year. ... If the material was from that time period, it would be a safeguards violation but not a violation of the 2015 Joint Comprehensive Plan of Action (JCPOA), which regulates nuclear activity from 2016. The green salt project was halted in 2004, and while all the documentation was carefully preserved ... there has been no indication of it having been resumed"
Your second article is from 2025 and it probably refers to last couple of years.. The US withdrew in 2018... Of course they continued enrichment after that withdrawal.
Let me add a bit more:
"The International Atomic Energy Agency (IAEA) released its quarterly report on Iran’s nuclear program June 6 [2018], and, unsurprisingly, the report found that Iran is complying with its commitments under the multilateral deal known as the Joint Comprehensive Plan of Action (JCPOA)." [1]
Again. You are wrong on this one! Iran adhered to JCPOA. US pulled out. Iran continued enrichment beyond limits defined by JCPOA as the agreement was dead by then.
> between 2009 and 2018, it said." -> this is smack in the middle of the JCPOA period.
The US withdrew in the 2018, so it is actually not "smack in the middle".
> And yes, this is from 2025, but it's about non-compliance during the period where the JCPOA was active.
It is actually not. You are not reading the material you are providing.
> The findings in the "comprehensive" ... pave the way for a push by the United States, Britain, France and Germany for the board to declare Iran in violation of its non-proliferation obligations.
> It would be the first time in almost 20 years Iran has formally been found in non-compliance.
Please read that last quote one more time.
> It would be the first time in almost 20 years Iran has formally been found in non-compliance.
But also this is about "violation of its non-proliferation obligations" not JCPOA.
You are going against the IAEA and US intel community which are both in agreement that Iran was compliant during that period. I think you have biases for which you are misinterpreting the facts. Either that or you are purposely spreading misinformation. In any case I will not purse this thread anymore.
"The finale of the PMD controversy has been a long time coming. In November 2011, IAEA Director General Yukiya Amano issued a detailed report — based on “overall credible” information from a “wide variety of independent sources” and the Agency’s own investigations — which concluded that, at least until 2003 and possibly beyond, “Iran has carried out activities relevant to the development of a nuclear explosive device.”
In the years following the report, the IAEA actively sought to gain a better understanding of those activities, but its efforts were stymied by Iranian stonewalling and obfuscation. Tehran repeatedly claimed that the evidence on which the IAEA was relying was fabricated and based on forgeries. It denied that Iran was ever interested in nuclear weapons or that it had engaged in nuclear weapons-related research and experimentation."
...
"This may come as an unpleasant surprise to American observers, many of whom probably assumed that sanctions relief would depend on Iran credibly disclosing its past activities, not simply fulfilling the undemanding, largely procedural requirements of the “roadmap.” Critics can be expected to attack the JCPOA anew for permitting sanctions to be relieved despite the December IAEA report having concluded that Iran has not made a full accounting of its past nuclear work"
So terrible agreement and Iran not acting in good faith. And we can debate technicalities and I'll even acknowledge that "technically" you're right but it's irrelevant.
So interrogating Gemini further to clarify the ground truth about the 2016-2019 period gets me to:
The Final Verdict
So, was Iran in compliance? Under the strict text of the JCPOA (2016–2019): Yes. They met the mathematical limits on active enrichment, which is why the UN, the IAEA, and the US State Department repeatedly certified their compliance during that period.
Under the spirit of the deal and international law: No. The premise of the JCPOA was that Iran had to come clean so the IAEA had a baseline to measure against. By hiding the Atomic Archive and keeping secret contaminated sites on standby, Iran proved they negotiated the deal in bad faith and violated their foundational NPT Safeguards.
I can live with that. So if you want to be "technical" sure. Either the agreement was bad and was upheld or it was good and was violated. Either way, Iran has acted in bad faith is the bottom line.
I will add that we don't have evidence that Iran was enriching Uranium in those secret sites during this period (one could even say we have some evidence they weren't). But that still doesn't change that they acted in bad faith and/or the agreement was bad.
Since my other reply was flagged, and I'm past the edit window, and I learnt a little more about the nuance:
- Technically Iran was considered to be meeting the requirements of the JCPOA during the 2016-2018 period in reports issued at the time.
- Iran failed to declare all its sites and programs before entering the JCPOA. This is known now, after the fact.
- Technically some argue that because Iran participated in meetings and filed papers they met the PMD requirements which were the preliminary requirement for the JCPOA to take effect. The nuance here is whether they technically fulfilled the requirements despite lying and hiding and then "only" violated the NPT or whether they violated the PMD.
- That Iran hid sites, material and equipment came into light after the Mossad stole Iran's nuclear archive. This is fact and was confirmed by IAEA inspections despite Iran's attempts to prevent that.
- When the IAEA asked to inspect those sites Iran engaged in a cover up operation and delayed access. After the sites were inspected there was evidence of nuclear material made by human activities.
- That material discovered by IAEA was not farther enriched which the supporters of the agreement claim is evidence that Iran didn't enrich more material. In reality Iran lied and hid facilities and so despite the samples taken by the IAEA not finding evidence of more enrichment the basic fact is that Iran acted in bad faith and so we just don't know. Maybe they only hid sites, equipment, and nuclear material but did not pursue further enrichment during this period. Maybe they did in other sites.
- Officially Iran was never found to be in violation of the JCPOA.
- The JCPOA was set to expire in October 18, 2025 after which there would have been no restriction on Iran anyways. That's another part of the argument that this was a bad agreement.
While it’s difficult to say to what extent they were going beyond there agreement, it’s clear that they were. I’m not aware of any evidence that it was to the level of, “they’re continuing to make quick progress towards a bomb.” Which is what happened when the US decided to reneg.
There were another seven years to negotiate what’s next and real progress made from both sides trusting each other. That’s the type of momentum needed for further diplomacy (e.g. counteracting more bellicose members of the IRGC.) Instead, we got the opposite. And for what?
There is a real divergence of experiences. That is one factor. There are people who are domain experts in some narrow complex domain who do such complex work that LLMs are still not there. I see some of that in my work place with large complex, domain specific languages, etc.
This is also challenging people's view of themselves as craftsman and the "crafting" of software. Something like carpenters who disavow power tools.
There is the worry about slop which is also real. I.e. that AI can and does generate garbage that ends up making things worse.
Worries about job security, the future of the industry, people's economic future and place in a new world where parts of their job get automated away.
I agree with you. Users don't care how the code is generated. This is purely economics there is no big market for "craft code" (like craft beer). There is only market for working software. And yes, LLMs are non-deterministic token generators, but so are humans, and LLMs are mind blowing. We live in the future. They don't replace software engineers quite yet- they are power tools.
That said the unevenness of contribution isn't strictly a large company phenomena. Small companies have the same uneven distribution. I've worked at two startups with about 4 people total and people were not equally productive.
That said, this is not necessarily the goal and productivity is also very hard to measure. It's doubly hard to measure across different types of work. One person can code up a greenfield back-end for something in 3 days while another can spend a week fixing some elusive infrastructure problem.
Not everyone is as good at everything. So we do have engineers who truly are much better than average. And in large companies most are average. But that is just one factor here.
1:1's can add value or they can not add value. Large companies can't just be flat so someone needs to manage people. A good manager adds value, a bad manager might subtract value, but that's orthogonal.
Is it demoralizing to work for a big and inefficient company? Sure. Is it more demoralizing for people who are motivated to get things done and are good at it? I think so. Go start your own company?
reply