trunnell

Netflix was a silicon valley start-up with a tech founder (Reed) who teamed up with an LA movie buff (Ted). They tried to solve a problem: it was too hard to watch movies at home, and Hollywood seemed to hate new tech. The movie industry titans alternated between fighting Netflix and making deals. They fought Netflix's ability to bulk purchase and rent out DVDs. Later, they lobbed insults even while taking Netflix's money for content licensing. Here's Jeff Bewkes, CEO of Time Warner, in 2010:

"It’s a little bit like, is the Albanian army going to take over the world? I don’t think so." [1]

Remember: this was the same movie industry that gave us the MPAA and the DMCA. They were trying to ensure the internet, and new tech in general, had zero impact on them. Streaming movies and TV probably wouldn't exist if Netflix had not forced the issue.

Netflix buying HBO is significant, but also just another chapter in this story of Netflix's internet distribution model out-competing the Hollywood incumbents. Even now in 2025, at least 12 years after it was perfectly clear that streaming direct to the consumer would be the future, the industry is still struggling to turn the corner. Instead, they're selling themselves to Netflix.

I was at Netflix 2009-2019. It was shocking how easily our little "Albanian army" overthrew the empire. Our opponents barely fought back, and when they did, they were often incompetent with tech. To me, this is a story about how competent tech carried the day.

Netflix has been rapidly buying and building studio capacity for a decade now. Adding the WB studio production capacity is a huge win for Netflix. It makes those studios more productive: each day of content production is now worth more when distributed via Netflix's global platform.

Same with WB and HBO catalog and IP: it's worth more when its available to Netflix's approx 300 million members. Netflix can make new TV and films based on that IP, and it will be worth more than if it was only on HBO's platforms.

[1] https://www.nytimes.com/2010/12/13/business/media/13bewkes.h...

If one your past managers did something recommended in this article but it caused problems, that's ok! It just means you have seen another failure mode that the author didn't experience.

I remember being in a meeting with a bunch of the best managers at a former company. "Why did you originally want to be a manager?" was one of the first questions passed around the circle. The most common answer was, "I had this one really bad manager and I figured that surely I could do better."

You can be confident there is a non-zero rate of errors and defects in any complex service that's moving as fast as the frontier model providers!

Maybe. What would you rather have?

A) rock solid Sonnet 4 with Sonnet 5, say, next April

B) buggy Sonnet 4 with Sonnet 5, say, next January

Seems like different customers would have a range of preferences.

This must be one of the questions facing the team at Anthropic: what proportion of effort should go towards quality vs. velocity?

They recently resolved two bugs affecting model quality, one of which was in production Aug 5-Sep 4. They also wrote:

  Importantly, we never intentionally degrade model quality as a result of demand or other factors, and the issues mentioned above stem from unrelated bugs. 

One could argue that batteries will have a bigger impact than solar. Batteries obviously let you decouple power generation and consumption, shifting anytime production to peak-time demand.

Less obvious is that local demand can fluctuate 2x. It usually dips mid-day and peaks 5-9pm (see the charts at www.caiso.com) when people come home and turn on their lights, oven, appliances, etc. This pattern happens throughout the year.

So forget solar for a moment; the ability to shift energy that was produced mid-day (even by a natural gas plant) to the evening would allow you to build fewer power plants. Nuclear + batteries might also be a good combination. Batteries get you closer to being able to solve for "average demand" rather than "peak demand."

This has nothing to do w/ California. California is just on the leading edge of battery installation. Solar just exacerbates the issue of the peak-to-trough ratio (evening vs. mid-day demand) due to mid-day solar "overproduction" causing it to be uneconomical to run gas plants mid-day. But solving for "peak demand" is still a problem in the absence of solar.

Still: most of the complaints about solar are answered when paired with large battery systems.

Yesterday evening's peak demand was between 7-8pm at 30.7 gigawatts. Supply breakdown around 8pm:

  Batteries: 8.4 GW
  Natural gas: 6.0 GW
  Renewables: 5.4 GW
  Large hydro: 4.4 GW
  Imports: 4.1 GW 
  Nuclear: 2.3 GW

“Digital-first production” can mean lots of things, so when you say Disney, Paramount, etc did this a decade ago, you’ll have to be more specific. Do you mean an end to end digital process? That’s not what this is about. Have you worked on a Netflix production? It’s night and day different from the studios you mentioned.

More interesting to me is who had the motive and opportunity, and this podcast makes a compelling case for who was behind it.

https://podcasts.apple.com/us/podcast/who-killed-jfk/id17146...

There’s actually a mountain of evidence pointing to a horrifying conclusion.

It’s not easy to summarize, and anyhow I don’t want to give spoilers here.

But I predict this won’t do what the authors claim: it won’t fix anything, and it won’t change technical interviews that much. At most it might increase the number of companies that require remote candidates to fly in and interview in person to do the same practical coding problems they used to do online. Which would be the standard from just a few years ago, anyway.

Practical coding problems (also known as work simulation interviews) are the most effective and least biased tools available for evaluating candidates. Not every hiring manager does a good job using these tools, but if you see misuse, take it as a sign that the manager might not be good to work for.

I have trouble understanding the authors thinking here that anything would be worth the cost of this: it’s a trust-busting tool where people are deceiving their potential employer. Trust is foundational to everything, in work and in life. We need more trust in our society, not less.

> how do good leaders stay in the detail and run great companies at scale?

It's a relevant question not just for founders but for leaders at every level.

IMO, one test for a "good leader" is whether they are capable of doing the work 1 to 2 levels down into their teams. The more familiarity they have, the more they are able to hire, fire, and evaluate those people. After all, it's pretty hard to evaluate work in an unfamiliar domain. Paradoxically, though, good leaders do not contribute to that work directly. So how do they maintain their skills if they don't do the work?

Consider the case of a front-line engineering manager with IC engineer reports. A good one will know their team's codebase, know where it could be easily extended and have good intuition for the time required for any given feature idea. They know the difference between good and bad code. But they NEVER submit PRs, mostly because the maker schedule/manager schedule problem [1] forces a choice of doing only one type of work well. (Every new manager I've seen who wants to "spend 10-30% of their time writing code" will either fail to support that code or fail to support their team as a manager, when in a fast growing team or company.)

The solution for eng. managers is to have the codebase on their machine, be able to build and run it, and occasionally implement their own experiments or POCs. These things NEVER go to production. It's meant purely to maintain the manager's familiarity with the codebase and staying current with their team's output. (Hat tip to CW).

Note that we still don't have good labels for these behaviors. "Hands on" and "hands off" confuses the issue-- is the example above "hands on" or "hands off?" It's both and neither, because those aren't useful labels.

There are other solutions for leaders higher up the org chart. The article mentions skip level 1:1s and niche area deep dives both with the purpose of evaluating leadership effectiveness. When I did these, I'd always start with setting the same context: I have two goals for this meeting and one non-goal. I want to hear about what you're working on, what's going well and where the challenges are. I also want to answer any questions you have about what's going on elsewhere in the company. My non-goal is giving you specific direction, since that's always between you and your manager. I'm just here to gather and share information.

The role of a leader is to set goals, share context and ensure the right team is in place, hiring and firing as needed. They need to know what's going on from top to bottom in the teams they lead, and in order to hire effectively, they need to be capable of doing the work 1-2 levels below them. But they never actually contribute 1-2 levels down, because that would severely undermine the people they've delegated to.

I think this is why so many had a knee-jerk reaction to in PG's founder mode essay, where he implied that founders have a special ability to bypass management layers and contribute directly (in the Steve Jobs example). I've seen it happen, and it failed 100% of the time. 100%. After establishing some amount of managerial structure -- wild guess would be after 50+ total employees -- contributing directly several layers down into your team is a recipe for disaster. The puzzle is how to lead effectively without making that mistake.

[1] maker schedule/manager schedule https://www.paulgraham.com/makersschedule.html * Bonus behavior: good managers are sensitive to booking meetings with any of their team members who are on the maker schedule.

In every case I've seen with healthy leadership, you could just ask who the top people are. Up to about 200-person size groups, you could ask any leader (even the top group leader) and they could just tell you. Because they generally know who is doing what, how difficult the various things are and who is turning in consistently outstanding work. (If the leadership can't do this, they might not be paying close enough attention.)

Performance reviews can serve many different purposes, but I don't think this is one of them.

I was reacting to the idea that the "top execs" are generally not the people you'd want. There was an undertone in pg's description that seemed to me like, "SJ hated those stupid and ineffective managers so he bypassed them to invite the people doing the real work."

I've worked in places with some ineffective managers, and at bigger companies I've witnessed a few ladder-climbing sociopathic liars, so I get the sentiment.

I just think it's more useful to figure out how to remove those ineffective people, or better yet build a cultural immune system that rejects them, rather than bypass them.

* CEO/founder should engage directly at multiple levels rather than only interact with the company through their direct reports. (Same applies at every management level, btw)

* Delegation is good, and it should happen in proportion with trust.

* The dominant culture at many tech companies is flawed and sub-optimal.

Bad points:

* "Founders feel like they're being gaslit from both sides". The two supporting points could both be true: "VCs who haven't been founders themselves don't know how founders should run companies, and C-level execs, as a class, include some of the most skillful liars in the world." However, it does not follow that the only option left is "Steve Jobs Style."

* "an annual retreat for [...] the 100 most important people"... I have trouble envisioning an effective org chart with lots of people at the top who would not also be in the "top 100" list. If your department heads are not your most skilled operators, then... maybe that's a good problem to fix.

* Assuming that the skills and intuition that make a founder successful will 100% apply to the very different job of being the chief of a 2000-person tribe. We should not assume that.

On average, everyone has an equal chance of needing to learn something new to succeed in a new situation-- founders included. Don't let pg's founder flattery go to your head.

I agree with you, I think. pg's point was that "Steve Jobs Mode" is the opposite of founders letting their C-level execs roam free. I don't agree.

I think the improved model is "free roaming managers with lots of transparency and accountability."

"Free roaming" and "no accountability" are a recipe for disaster.

But a CEO/exec/manager who reaches 2, 3, or more levels into their org and gives specific direction is a recipe for mismanagement. It violates the golden rule: don't create a role with two bosses.

Sounds like chesky and pg want to turn the tide on that dominant culture in software companies. And I couldn't agree more! A big problem IMO is that most "professional software managers" are taught a management style that focuses on risk. Risk-aversion permeates every decision from compensation to project priorities. It's so pervasive it's like the air they breathe, they don't even realize their doing it. This is how things run in 99% of companies.

So, my fellow hackers. There is a better way. It's neither the Steve Jobs model nor the John Sculley model. Looks like pg has not yet found it. I hope he does, though. It would be great for YC to encourage experimentation here.

I have solar, a powerwall battery, a high efficiency heat pump, and... a poorly insulated 70-year old home in Silicon Valley.

It can be wildly expensive to properly mitigate poor insulation: you need good air sealing, insulation around the entire building envelope, double or triple paned windows, and a different HVAC setup with dedicated fresh air ventilation. In other words, it requires a major remodel in some cases. Homes need to be built with energy efficiency as a top concern, and I wonder sometimes if that is going to require re-training and incentivizing the entire construction industry. Fewer than 1 in 10 contractors I talked to even knew what I was talking about when I asked for how they would do my project. "What's an ERV?" is a common question I heard. Many still think that gaps are good because "a house needs to breathe."

I worked with this team (and its predecessors) during my time at Netflix. They achieved several "holy grails" of video encoding: a perceptual quality metric (VMAF), optimal bitrate selection per 2 second video chunk, and then optimal video chunking to be scene based rather than a fixed 2 seconds. Doing any of that in a research lab would be a challenge, but pulling it off at Netflix's scale is epic.

You might need some background on how adaptive video streaming works to fully grok this article.

But this is also just a story about a massive refactoring of a large, critical system. How many companies have you worked for that aggressively pursued refactoring/re-engineering their central systems? At most other places, I've seen risk aversion, fear, and mismanagement conspire to kill innovation. Not so at Netflix.

But you can also just enjoy the story of developer achievement!

1) Price seems too low for unlimited LLM usage and there's no monthly fee... so maybe you are the product?

2) Website has no information on the people behind this.

3) In the keynote, the demo of logging into to other services like Spotify appears to just be stealing the auth tokens from the laptop and shipping them to the R1 device. Not a good sign.

4) The founder's voice-over insists that "we value privacy" and "we do not hack" and "we do not create fake users." Protests too much?

5) They're not really addressing the trust-building necessary to convince people to use this like a personal assistant. It's waaaay too opaque.

Neat product, though. Maybe they're just launching super early, nothing is figured out yet, the website isn't done, etc.

  In a statement to CNBC, Microsoft said the ChatGPT temporary blockage was a mistake resulting from a test of systems for large language models.

  “We were testing endpoint control systems for LLMs and inadvertently turned them on for all employees,” a spokesperson said. “We restored service shortly after we identified our error. As we have said previously, we encourage employees and customers to use services like Bing Chat Enterprise and ChatGPT Enterprise that come with greater levels of privacy and security protections.”

When the user and developer are the same people, the cost of finding the requirements can sometimes be nearly zero. In this case, that person will themselves choose the best tool for them. And the development cost will be very cheap -- for example, think of a short script you wrote to do something just for you.

But when the user and developer are different people, much care should be taken to avoid cost explosion. To the outside this can look like "coding is expensive!" which is kinda true, but really it's the requirements discovery that dominates the cost. People don't understand each other perfectly, can sometimes talk past each other, etc. Iterative development's purpose is to mitigate the problem of building for the wrong requirements.

No-code tools can unblock "developers" who own the requirements but don't have programming skills. That might make those projects very cheap. Or it could do the opposite: if the non-programmers don't know think carefully about what they're building, then they might end up spending just as much time as programmers would have as they iterate towards a complete solution.

After a bit of searching, it's unclear what degree of "patent risk" comes with JPEG XL. JPEG historically was subject to patent troll lawsuits until the patent expired in 2006.

Please note that it's not enough for there to be a "royalty-free reference implementation" of JPEG XL, even if it's licensed with Apache 2.0, because you can't be sure from a glance that the Apache license patent grant includes all relevant patents. If you care about open source and free formats, you should look for two things: a comprehensive patent pool transferred to the standards body AND a royalty-free patent license to anyone with no strings attached.

The game here is that companies with potential claims over some techniques used within codecs have an incentive to withhold their patents from the official pool until years after adoption. Then they sue the biggest users of the codec (like Google) for obscene sums of money. That's why ALL of the patents used in a codec must be assigned to the standards body for open licensing, and you have to be SURE that none are withheld. This is difficult.

AVIF (and it's standards body, the AOM) was created in part (I believe) to solve this very problem. All the major tech companies are members and they've effectively agreed to a patent truce with regards to codecs.

This is arguably the most important commercial concern in distributing a browser for free that includes codecs. If you ship unlicensed codecs, some random company can crawl out of the woodwork 5 year later and sue you for a billion dollars.

In my view, AVIF only needs to be competitive with compression and quality. It patent risk is so low that it is the obvious choice. AVIF is truly open, there are multiple implementations, and its reason for existing is to solve the codec patent problem.

Source: I was near the activities within Netflix that helped found the AOM.

Disclaimer: I'm not a lawyer and this isn't legal advice; also I'm several years out of date w.r.t. JPEG XL specifically, so I'd be happy to be corrected about the relevant patent risk. Maybe someone has better info?