steveBK123

I think AI is creating some of the same problems in non-coding tasks that it has created in coding.

That is - you can do more & have faster iteration loops, but you can basically burn all the gains on stupid stuff no one needs.

If your org likes to scramble, and/or your management has ADHD.. then LLM slop cannons pointed at things like dashboards, status reports and decks is extremely exciting. People are throwing together materials about content they don't understand for people who aren't going to read it. But the motion feels incredible.

Q&A at all hands meetings is a trap to fire people who don't drink enough kool-aid.

Do not, under any circumstances ever ask a question other than "gee golly sir, you sure are smart, how do I grow up to be like you".

Completely opposite for me. Have mostly non-tech family across a variety of industries. Finance, accounting, education, pharmaceuticals.

Everyone got the "use this or we will kill you" speech from management, while finding its actual usefulness (they aren't coders and weren't using it for code!) to be questionable vs the hype.

Further a lot of people got the "you better AI-max or-else" ultimatums in Q1 only to get the "you idiots why are you using so much AI this costs money" by Q2/Q3 whiplash.

> This is the first time ever a technological advance is so cordially dislike

Lot of contributors to this.

Macroeconomics are bad, high-ish inflation and hard job switching market.

Meanwhile everyone's bosses telling them to use AI or they will fire you, AI lab heads bragging/threatening the replacement of all knowledge work in next 18 months (unless AI kills us all which they also scaremonger about).

So use it until it replaces or kills you. Good luck staying poor because the IPO is coming after all the insiders 10000x their investment and the bag left for retail investors in the public market is physically impossible to 10x or 100x in most of our lifetimes (again if AI doesn't kill us and/or make us so poor we won't own stocks anyway).

So yeah, lots not to like!

I had one recently that was like.. I'd be better at this test if I knew less / was less experienced.

An hour to implement a toy version of a component which, the last time I did it "in real life" blew everyone away that I had a prototype ready in 1 day. That real life prototype went through weeks and weeks of iteration before it was even useful, and everyone was happy with that. But yeah that's something to time box to 1 hour in an interview environment.

So turns into an exercise of - which of the 100 pitfalls/nuances do they want me to pretend don't exist in order to complete it in 1 hour. Opinions vary!

There were multiple questions of this form.

One could pass this test blindly as a new grad not knowing the 100 pitfalls of actually building one in real life.

More often than not the test is at best tangentially related to the role.

At worst it’s the biggest hardos on the team writing a bunch of brain teasers and gotchas as a hazing ritual.

The more inhumane the screener the worse the job in my experience.

Had one recently that waived me thru and then sent me straight to talking to senior management/stakeholders/etc but literally no one on the team itself.

Couldn’t get a straight answer as to what team does, what role is for, real day to day I’d be walking.

Only clear thing was scale of architectural mess stemming from a political mess… and not a lot of interest in changing it.

They wanted a seasoned tech body to throw into the grinder 12-15hr days to replace the last one. No thanks.

Pretty funny that tech makes candidates jump through all sorts of automated, inhumane loops to even advance to talking to a real human in the hiring process.

And now you get results like this.

> The article then ends before saying if the new setup solved their problems.

They got to put ClickHouse, DuckDB and Trino on their resume & LinkedIn. Problem solved!

Yes but those convertible notes only exist to grant them access to compute they would otherwise need to pay fair market value for.

If I drive a taxi making $200/day and need $150/day of gas.. but convince Shell to take a convertible note instead of $150.. Do I make a) $200/day profit, b) $50/day profit, or c) $200/day - (change_in_value[convertible_note]) in profit?

Some are arguing that C is somehow fake, but its obviously more real than A!

Well they can’t have it both ways on the accounting front.

These complex securities they took a mark on only exist as a way for them to access compute at deep deep discount by issuing suppliers lottery tickets.

Post IPO when they don’t have lottery tickets to issue as easily the numbers should be interesting.

In some ways, the biggest "new form of business/commerce" that ubiquitous smartphones lead to was the entire gig economy.

It's not even clear that the "frontier AI lab" is going to be the best, most profitable business model to evolve out of this tech. Especially if models are easy to distill, a new lab can start on 3rd base, and China open weights aren't far behind.

The winning business model may be 1 or 2 abstractions removed from "build a really good LLM model", and yet to be revealed in the marketplace.

> B) nobody is IPOing at $400 million like Amazon did

Exactly You can't wait to IPO until you're already one of the 10 biggest companies ever AND also grow 2500x post-IPO.

Retail is now exit liquidity for Big VC.

None of the current vintage of IPOs are going to 2500x in our lifetimes (inflation adjusted at least..). The VC to IPO process is totally different now.

Companies are staying private longer, raising many orders of magnitude more private funding, and IPOing straight into "top 10 biggest companies in history" size.

If you IPO at $1.5T you are already equivalent to ~1.5% of World GDP.

Thus is just for scale, as comparing GDP vs a company market cap doesn't make total sense since GDP is 1 year economic activity and market caps represent discounted future cashflows of future economic activity...

Let's say Anthropic IPOs at $1.5T and grows even 1000x over 20 years, it would be worth $1.5P (petabillion?). If it had a 20x earnings multiple like google this would imply Net Income of $75T. Mag7 total Net Income for 2025 was $500B or $0.5B.

World GDP is $125T, and has grown less than 5%/year before/thru/after DotCom. So lets say in 20 years its $250B and Anthropic is capturing 30% of world GDP in earnings (not revenue, that would be even higher)!!

So the math doesn't math.

My proposal has always been that we adjust the length of an hour so that the sun always rises/sets at the same time. /s

Maybe author means we should actually enforce our antitrust laws that are on the books, on their merits.

Rather than the multi-decade consensus "if it lowers consumer prices in short term its ok" or 2024+ "if they donate to my ballroom, they get a pass".

Right it started with SaaS and content subscriptions. We also have all the gamification, enshittification and up-selling in the physical-world Uber-For-X business models.

They've made all the delivery slower and added an option for you to pay $5 more to make it arrive at the speed it used to. We've jacked up fees and added a monthly subscription that reduces some of those fees back down.

As they take a larger cut from restaurants, many have increased in-app prices so you are paying multiple premiums&fees.

First non-cherry picked example I checked first restaurant in-app: $16 dish is $19.20 in-app , $5 delivery fee (flat-ish per order), $2 marketplace fee (scales with order size) to make it $26.80 pre-tax. +70% increase on direct retail in-restaurant price.

According to cursory search:

* Spain HSR ~45M annual journeys / 30B annual passenger-kms

* Italy HSR ~65M annual journeys / 50B annual passenger-kms

* France HSR ~150M annual journeys / 110B annual passenger-kms

"Italy’s annual high-speed rail (HSR) passenger fatality rate is effectively zero, as there has never been a passenger fatality recorded on the country’s dedicated high-speed network"

"France’s dedicated high-speed lines (LGV—Lignes à Grande Vitesse) have historically maintained an extraordinary safety record. Across several decades of commercial operation since launching in 1981, there has never been a passenger fatality resulting from a high-speed crash during commercial service on a dedicated high-speed line"

Spain by comparison has 2 fatal accidents adding up to ~100 passenger deaths this year alone. Something like 300 fatalities in its history.

This is quite bad.

I'm still running a 4 year old M1 Max 32GB Mac Studio as my daily and really haven't found anything that it shows sign of slowness/age with.

That said I don't do any local AI or 8K video editing type stuff.

On upgrade I could probably live on a moderate/high spec Mac mini since the only reason I went Studio originally was the resolution/number of displays it could support.

At this point, Spain must have the worst high speed rail safety record in Europe if not.. more broadly?

I am a little surprised doing some light searching that its one of the longest high speed rail systems / highest passenger count HSR systems in Europe?

I've only taken HSR in France & Italy and not been to Spain in over 15 years so was not aware of the build out.

You don't even need there to be any kernel of reality statistically.

If a model is trained on data that shows some bias due to prior human decisions, it would continue to follow those biases while being a lot harder to prove/sue. There were examples of this with AI assisted resume screening & criminal sentencing for example.

A lot of systems are either adversarial or the v0 agents are relying on a sort of free rider / human good will that won't scale when 100M people have agent bots.

Apples iPhone call answering for unknown numbers was OK until all the overseas robocallers made their robot respond somewhat conversationally and hang on the line so that it vaguely looks like a real enough call. I can still tell its a robot since I am not senile yet, but I still have all the notifications about the call/transcription/voicemail/etc to tap.

People deploying agents to call restaurants for reservations works for like the first 10 calls that restaurant gets and now they are going to hang up on everyone's bots. Many websites already blocking agents. Some websites where agents were scraping for reservations/inventory are outright locking peoples accounts. Etc.

https://www.businessinsider.com/resy-suspends-vcs-account-ai...

https://www.cnbc.com/2026/10/13/jpmorgan-ai-agents-account-a...

https://finance.yahoo.com/technology/ai/articles/metas-muse-...

It speaks to a lack of understanding of the economy as a complex adaptive system.

Same idea with all these agents that are going to negotiate our insurance down, haggle with our cell phone providers, and find the best hotel deals.

Your counterparty will have its own agent, a larger budget for that agent, and a lot more skin in the game to ensure they retain their advantage (cell phone bills are 100% of the cell companies income but is only 0.5% of your spending). Good luck.

Sigma is the last great experimenter in the industry and I appreciate everything they do however wacky (former fp, fpL and various i-series lens owner)

That sounds like being an actual founder though in which case downside is large. 0 (or negative) TC because you have no profit and you are bootstrapped.

A lot of intensity is just people in charge having status seeking goals that were formed at a young age.

It may surprise people here, but there are teenagers who set out with 10-20 year goals like "I want to found a startup / be a CEO / have my own venture fund / etc" and then make every educational/career/relationship/lifestyle decision in that context.

For most people we want to work on interesting projects that pay us enough to live the lifestyle we can enjoy.

I think this is why a lot of successful founder types come from upper middle class backgrounds. If your dad is a mechanic then "being good with computers and going to college" is an aspirational goal.

If your parents were first in their family to go to college then "go to a nationally known college and working at a Fortune 500" is an aspirational goal. If your dad went to an Ivy and runs his own business then the aspirational goals you have instilled in you are much higher.

I've done a few types of jobs myself and unfortunately not really found offer frameworks that look like "capped downside and uncapped(ish) upside" based on performance.

Ideally what many senior engineers are probably looking for is - I can live on 50-75% of my TC a few years without setting myself too far back, and I'd work really hard on an interesting project that has a >25% chance of rewarding me 2-4xTC. It is not too expensive to roll that dice, and I have a reasonable chance of hitting in one career.

Instead we get a lot of lottery ticket profile things that are like "1% chance of 1000xTC". The EV looks good, but you don't get enough rolls of the dice for that 1% to pay off in a single lifetime let along career.

> * What are the downsides of turning up the knob?

I once worked at a relatively small (<1000) person firm that basically had a near extinction event due to failing to fund a secondary datacenter / sensible data backup / business continuity strategy for years (despite engineering raising the alarm).

A team of 20 engineers worked 16 hour days for 2 weeks through the weekend to rescue them from their own decisions.

At the end of it they had a 10 minute zoom with the team to thank us and tell us they were going to - send us a ($10) bottle of wine, and grant "2 days off in lieu" (we were already on "unlimited vacation" plan) for the extra ~14 days of work.

That was a single-use only knob due to their behaviour.

No one cared enough again to do anything above & beyond, morale went to zero, and senior members of the team had all left within a year. The thank-you was so bad it was probably better that they not have done one at all.

Absolutely survivorship bias, as well as - intensity on the right things / right direction / level of decision making are important.

Go yell at your C-suite / directors with equity that have/will have F-you money from the success of the business. Don't think your ICs 3 levels down the hierarchy making 1-3x inflation-level raises will be as driven.