Instead of doing nonsense to waste tokens yourself you can put nonsense in the AGENTS.md and have the whole team waste tokens. @ a bunch of files to add them to the context when they don't need to be. Add instructions to have the LLM do more tool calls, maybe add some verbose flags.

It's less accurate than copy/pasting shit off Stack Overflow, the thing we used to say was the sign of a bad programmer. Everyone using it is miserable, and nobody using it has actually delivered the results they keep saying they're getting (it's been 6 months since you all said you're 100x more effective, where's the 50 years worth of new software, AI bros?)

I dunno that wasn't my experience in the last 6 months or so. 2 years ago yeah it was basically this but nowadays I can get the thing to do some medium complexity work on our codebase and it's pretty much gonna get it right like 95% of the time with maybe some tiny code quality reduction in the form of some duplicated code or such but the thing is pretty much going to work.

Also I built a personal project with it that I wanted to do for a while but never got around to it and it pretty much did it as I wanted it. Granted it was a lot of back and forth and lots of debugging/testing passes but in the end it actually works. Not sure what the overall code quality is but the overall project structure makes sense and the little code I did look seemed solid enough.

I would say the people on my team are about 3x more productive than before. Of course, they immediately cut our team by 2/3rds, so...

It just lays more bare what the people up top care about. Most devs eventually learn this but they are often somewhat insulated as they are left to their own devices to build a product via a given framework or standard. But the C-suite has never cared about code quality or security unless a big customer or investor cares about it. The C-suite is focused on making money, especially through disciplining labor. They are very happy with the worst thing you've ever seen so long as it appears to function and neither big customers nor investors are complaining about it. Tech is usually treated as a cost center, it is something to minimize and fulfills a purpose that they would rather contract out.

These are folks with big egos and little education so they also think they know enough about the topic to manipulate their workers and discipline as needed. This has never been a problem for them before except when it comes to tech and how much they have to pay their workers, as there is limited supply. They are now having a great time attempting discipline that was otherwise difficult for them to achieve, so this is a double-whammy of standard labor discipline plus scratching a long-held itch in profit maximization. It doesn't even have to work, it just has to fulfill those basic criteria from earlier.

In large part this is because there is a substantial disconnect between production and profit, the actual product being sold and how money is made is often two or three degrees removed from "we made $x sales and had $y costs" that the C-suite can just tank four companies in a row and still make out like bandits - so long as they please the big customers and investors.

Your boss is probably clueless enough that even explicit written frustration will not be read as such if they don't have a bone to pick with you but even the most charming writing could be insubordination if they want to see it.

I think that's one of the things that irks me most about this LLM bubble. I think ML has really cool applications in research, approximate tasks with margins for error like image classification, upscaling etc. It was one of my favorite topics during my CS degree. But now it's tainted by the gen "AI" slop machines, so all the interesting breakthroughs of real practical ML get chalked up to "AI" innovation and people just assume chatGPT is responsible when it's actually a team of researchers building small efficient programs. Ofc ML is less efficient than traditional computational approaches that can work deterministically so those should always be prioritized, but it has its uses for sure.

For example, I want to know more about the role of agricultural cooperatives in China. I don't speak Chinese and have no experience with the Chinese internet or Chinese state and party websites and databases. How the fuck could I figure that out?

Well, I asked AI this:

I want to understand the scale of cooperative development in China. How many are there? What's the scale of their economic impact (property owned, production levels, etc)? How quickly has this sector grown? How many people are members? How does it prepare proportionally to other segments of the agricultural economy? How does China see these strategically in the process of building socialism?

And here was its response:

Excellent set of questions — this needs real data gathering. Let me run several searches, in Chinese where appropriate, since the official numbers live in Chinese-language sources.

Strong data so far. Let me dig deeper on the growth trajectory over time and the strategic/ideological framing:

One more quick search to anchor the comparison against overall agricultural output:

I now have a solid picture from official and semi-official Chinese sources. Let me synthesize across all five dimensions you asked about.

Snapshot: the scale of China's farmer cooperatives

These figures come from the Ministry of Agriculture and Rural Affairs (MARA) via the State Council Information Office (April 2025 press conference) and a MARA year-end summary published December 2024. By end-2024, there were 215.3万 registered cooperatives and roughly 400万 family farms in the national directory. Membership exceeded 100 million households according to a 2017 Xinhua report — over 46% of all farm households at that time — and MARA's 2024 data still describes co-ops as reaching "nearly half of farm households."

Growth trajectory: explosive rise, then consolidation

The single most striking fact about this sector is its speed. According to MARA expert Shao Ke, quoted in People's Daily, registered cooperatives grew from 26,400 in 2007 to over 2.2 million by 2021 — roughly an 80-fold increase in 15 years, all triggered by the 2007 Co-operative Law.

Sources: Xinhua (2017), MARA development reports (2019), People's Daily (2021), MARA/SCIO (2024). Note the shape of this curve: near-total flattening after 2019, and an actual decline by 2024. This is not a market correction so much as a deliberate policy pivot — Beijing launched a nationwide "shell co-op" (空壳社) cleanup campaign in 2019 to deregister cooperatives that existed only on paper, largely formed to capture subsidies. As one critical analysis by Professor He Xiurong of China Agricultural University notes, regulators estimated shell co-ops at 60%+ of registrations in some surveys, and Chinese scholarship openly debates whether "true cooperatives" in the classic sense are widespread at all. Official policy language has correspondingly shifted from 数量扩张 to 质量提升 since roughly 2019.

Economic impact

Hard numbers from MARA's December 2024 summary and People's Daily coverage of MARA data:

• Annual operating revenue of ¥630.92 billion (6309.2亿元), with additional service value to member households exceeding ¥880 billion
• Member benefits: per-member surplus redistribution averaging ¥1,471/year, plus an average of ¥15,000 per member in unified purchasing/sales services
• Land scale: grain cooperatives alone number ~540,000; the average cooperative holds 460 mu (about 30 hectares) of land-management rights contributed as equity
• Vertical integration: 104.3万 cooperatives provide integrated production-processing-sales services; 120,000 have founded processing/distribution/sales entities; 65,000 engage in e-commerce
• Quality signals: ~103,000 hold registered trademarks; 63,000 hold product quality certifications

To put revenue in perspective: this is gross operating revenue, whereas China's primary industry value-added was ¥9.14 trillion in 2024 (National Bureau of Statistics), with the broader agriculture-plus-related-industries figure at ¥20.6 trillion. So cooperatives' direct output is a meaningful but mid-single-digit percentage slice of the agricultural economy — their strategic weight lies less in output share than in organizational reach, as I'll get to below.

Proportional position within the agricultural economy

The comparative landscape (2024 data unless noted):

• Small farmers still dominate structurally: 98% of agricultural operating entities are smallholders, farming over 70% of arable land, with 2亿+ total operators (People's Daily, MARA)
• Family farms: ~4 million in the national directory — more numerous than co-ops but individually smaller; over half earn ¥100k–500k annually
• Leading enterprises (龙头企业): only ~87,000 (2018 NBS figure), but they anchor the "dragon-head enterprise + co-op + farmer" contract-farming chains
• Agrarian socialized services: arguably the fastest-growing segment — 109.4万 service entities providing 2.14 billion mu-service-passes annually, reaching over 94 million smallholder households, with co-ops themselves constituting a large share of providers

So the honest comparative picture: cooperatives are the organizational bridge in a sector where the production base remains overwhelmingly smallholder. Their seed-grain farms average 20% higher yields than the national mean (MARA), and their function as intermediaries — bundling smallholders into procurement, standards compliance, certification, and marketing channels — is what policy treats as their core value, not raw output.

The strategic framing: cooperatives in "building socialism"

This is where the ideological architecture becomes explicit, and I'll present it in the official framework's own terms alongside the analytical distinction:

The Marxist lineage is cited directly. Party-theoretical literature (Qiushi, CASS journal articles) traces the cooperative idea from Marx and Engels' concept of "cooperative production" under collective ownership as the path beyond poverty, through Mao-era collectivization — by end-1956, 75万 advanced cooperatives enrolled 96.3% of farm households; the people's communes (1958) took this to 99.1%. The commune system's failure is treated within China as the reason the post-1978 reform restored household farming under the "dual-tier management system" (统分结合双层经营体制).

Today's cooperatives are framed as the new vehicle for old goals. Xi Jinping's formulation, quoted across these sources: hold the political direction of rural revitalization, uphold collective land ownership, develop the "new collective economy" (新型集体经济), and "walk the road of common prosperity" (共同富裕). Note the theoretical nuance in the Shandong Academy of Social Sciences definition: the new collective economy includes not only traditional collective ownership but cooperative and shareholding-cooperative forms built on private property — labor joined with capital, technology, and management. So cooperatives are seen as an instrumentalist form: they embed peasant interests, prevent asset concentration ("精英俘获" — elite capture is explicitly warned against), and link smallholders to modern markets.

Concrete mechanisms connecting to socialist goals: the 联农带农 (link-farmers-bring-farmers) policy now ties state subsidies to demonstrable farmer-income linkage; "党建+合作社" (Party-building + cooperative) models, often village-Party-branch-led land-share co-ops like the Henan example in People's Daily, blend the collective and cooperative layers; and the 2021 co-op law provisions restricting internal credit cooperation to members only reflect wariness about financialization. The strategic endpoint in official discourse is 农业强国 — with co-ops as the organizational connective tissue that lets a smallholder-majority countryside participate in "Chinese-style modernization."

The analytical caveat I'd layer onto this: scholarly debate inside China is more candid than the policy documents — the He Xiurong piece notes research finding many registered co-ops are controlled by rural elites or companies rather than operating as member-controlled organizations, and that advanced economies' agricultural cooperatives have been shrinking for decades. The grand narrative of co-ops as the vehicle of common prosperity coexists with an unresolved empirical question about how genuinely cooperative the sector actually is.

Sure, there's a bit of needless slop and speculation at the end in particular, but this is obviously a useful tool for understanding things that are otherwise very hard to access.

I like the idea of using generative AI for getting visuals for (PRIVATE) TTRPGs

I like the idea of using my imagination