ZainRiz

I'd respectfully push back on the framing here.

If you look at value as purely the LLM output, then there's a valid argument that the best frontier models will always be better than fine tuned specialists. (I'm not convinced personally, but it's a defensible claim)

But that misses two dimensions: 1. The cost of acquiring that output 2. What is actually "good enough" for that specialist domain

Not every output needs to be the best to produce value.

And as specialist models increase in cost, their cost/value proposition goes down.

At some point, there's a threshold where cheaper, fine tuned models are "good enough" at the task and also substantially cheaper than the expert models.

That's where fine tuning helps.

Personally, I became a believer in fine tuning after fine tuning a 1B Qwen model as a second pass over my local voice transcription app, achieving excellent accuracy at ~zero token cost and waaaay lower latency than if I'd invoked my Claude subscription under the hood.

If you're in finance, you earn enough to live in NYC

> imagine what happens to NYC when all the people doing the menial work keeping the city alive (have to) move away? Whoops, now everyone is going to drown in trash and feces!

That's not a "the poor middle class folks in NYC need help" story, that's a "the rich folks of NYC need folks to serve them" story. They're welcome to strategize however they like to incentivize people working there. Manipulative heartstring tug are not welcome however.

Maybe now someone will finally fix the bug that causes claude code to randomly scroll up all the way to the top!

They also have a "keep going" keyword, literally just "continue" or "keep going", just for logging.

I've been using "resume" this whole time

This is so cool! Now, I didn't notice anything about waterproofing on your website.

Any plans to come out with a version I don't have to worry about getting wet?

Avoid casing entirely and use single letter function and variable names

Case insensitive, of course

Same here. I found the biggest beneficiary of a design doc was my own thinking process (Xoogler here)

Google has the concept of a mini-design doc called a "One-pager"

In 2018, the template for this "One-pager" was two pages long

With Microsoft they tell you years in advance when the product will reach EOL, so you have plenty of time to prepare

On top of that, most of their products are software running on your desktop and will still keep working after EOL

And if you're a government or enterprise with spectacularly big wallets, you can choose to pay Microsoft big $$ to keep shipping you additional security patches

I think the claim is that some other company in the robot vacuum business can buy iRobot instead of Amazon and get that sweet sweet IP

+1. At 22, working as a full time software engineer at a tech company, I still didn't think I knew how to do anything useful

Even though I've drawn pictures, built toys and contraptions contraptions out of wood, made a rube golderg machine

Still...

The thing I'm the proudest about...

Is that time I opened a jammed coaxial cable with nothing but two butter knives.

The challenge with books like The God Delusion is that it's barely one step above strawman arguments.

It attacks claims made by some simple lay people, based on their half-baked understanding of their own faiths (which they never really dug into). However, people well versed in the religion don't actually make any such claims.

The arguments being debunked are ones that religious scholars themselves would have disagreed with. Such misrepresentations don't guide anyone

Kindly also report the usage of "Chai Tea" ('Chai' means 'Tea' in Pakistan/India, which is where that flavor originated from)

There's two aspects to value creation:

1. How much more value you can now create. E.g Moving from assembly to managed code was also a 1000x shift

2. How much of that value you actually capture, as opposed to it going to your CEO, customers, etc

The people in developing countries have the misfortune of being able to capture only a minuscule fraction of the value they create

> but there is no known island within that lake.

brb, I'm gonna go toss a big rock

No need to make every person categorize/rank their own content. But let them choose which of the 3rd party ranking algorithms they'd like to have serving them their feed today.

Good point. It doesn't necessarily have to be a paid membership, but it would need some type of reputation formation. Ability to pay money is one way to grant you a "prob not a bot" reputation.

There's many other ways to set reputation thought:

- I've seen private online communities where you need one or two members to "vouch" for you

- StackOverflow has it's own reputation model based on how helpful you've been in the past

- Bitcoin et-all use proof-of-work as a reputation model.

I'm sure there are many more just waiting to be discovered

The tools for a community to self-moderate is the key here, along with communities being able to define their own boundaries.

Reddit communities with active moderators are often great places to have respectful conversations. Twitter, on the other extreme, has virtually zero moderation tools for any group. You get to live with whatever rules Twitter sets for the entire platform (and die by the group that's able to hack them)

Trust-less information was the norm for most of history.

The past century has really been an anomaly when it came to trusting news, with photos and videos seeming to be reliable proof. Before that, societies had to be structured very differently to account for the lack of proof. You'd hear people from out of town talking about what they saw in a neighboring city.

You'd need to judge how trustworthy the person was.

People would expect a chain of narration, to understand how _that_ person came to learn a bit of information (or if they claimed to witness it first hand).

As AI generated content becomes more popular, I predict we as a society are going to go back to relying more and more on the reputation of the speakers in question.

Who might you consider reputable? People you've met personally, people who your community respects, and of course, influencers who you follow who's persuasive words match your pre-existing world views. Perhaps something else will emerge

Found this while exploring pytest's cache. For something so simple, they put a lot of thought into the spec

The past century has really been an anomaly when it came to trusting news, with photos and videos seeming to be reliable proof. Before that, societies had to be structured very differently to account for the lack of proof.

You'd hear people from out of town talking about what they saw in a neighboring city. You'd need to judge how trustworthy the person was. People would expect a chain of narration, to understand how _that_ person came to learn a bit of information (or if they claimed to witness it first hand).

As AI generated content becomes more popular, I predict we as a society are going to go back to relying more and more on the reputation of the speakers in question.

Who might you consider reputable? People you've met personally, people who your community respects, and of course, influencers who you follow who's persuasive words match your pre-existing world views

I'm surprised by the combination of "Writing a book + public speaking opportunities + (apparently public) twitter account" with not wanting to dox yourself in the blog post.

Do you have two+ very different online personas? I can't imagine you doing public speaking without your book readers/twitter followers knowing exactly who you are. And if they know, then why keep your blog covert?

I'm really curious about what content you post publicly vs anonymously.

Is the claim here that the city's spending needs went up much faster than inflation? If so, why?

First I'm hearing about nerfed iphones, but if true this sounds like a reasonable decision from a business perspective.

They wouldn't want customers worrying about whether they got the better qualcomm iPhone X or the slower intel iPhone X.

Apple could have chosen to make everything with the faster qualcomm chips, but they prob wanted to derisk their supply chain by diversifying onto two chip manufacturers (and perhaps play them against each other for negotiating power)

Either ways, once you arrive a the result that you need to chip manufacturers making chips for the same phones, you need to make sure your customers have practically the same experience regardless of the underlying hardware they end up with

Consistency is key

> Over 50 percent of Black children in the U.S. will experience a child welfare investigation before their eighteenth birthday (nearly double the rate of white children). Nearly 10 percent of Black children will be removed from their parents and placed into foster care (double the rate of white children)

While the numbers above sound horrendous (and they really are!) I wish they normalized the data to only consider poor households. That would give a much better picture of how much of the existing system is biased against a given race vs being biased against poor people in general.

Today I realized how xkcd prob got it's name

Wow, these games would be impressive for a college kid to be making!

"problem" is a bit of a strong word, but when someone says "Cancer Death Rate Dropped" dropping people assume "Cancer Rate Dropped", which is not at all the same thing

The article provides no numbers about cancer rate dropping. Yet the false association between the two ideas is so strong that even the authors themselves seem to theorize that the cancer rate itself is dropping because people don't smoke as often now.

Conflating those two concepts seems like a Very Bad Idea