Ad Infinitum · Matthias Ott:

One approach, proposed by Google Research, is what you might call a “token auction.” In this model, advertisers don’t buy ad slots on a page. Instead, they bid, token by token, on the actual text the model generates. Each advertiser brings their own LLM, and an auction mechanism decides whose model gets to influence the next word. The output is a weighted blend of competing interests, shaped by who’s willing to pay more.

Another approach – also from Google researchers – fits the new “Search” much more precisely. It’s called “prominence allocation.” Here, when a user submits a query with commercial intent, the system runs an auction that doesn’t just decide which ads appear, but how prominently the LLM writes about each one. The auction outputs a prominence score for each advertiser, essentially telling the model: give this product 35 words, that one 20, and this one zero. The ad isn’t next to the answer. The ad is the answer. Or rather, it shapes how much space and enthusiasm each product gets within the answer.

This is terrible. I’m glad I signed up for Kagi a while back.

It’s Time to Dismantle the Technopoly | The New Yorker:

In a world where a tool like TikTok can, seemingly out of nowhere, suddenly convince untold thousands of users that maybe Osama bin Laden wasn’t so bad, or in which new A.I. models can, in the span of only a year, introduce a distressingly human-like intelligence into the daily lives of millions, we have no other reasonable choice but to reassert autonomy over the role of technology in shaping our shared story. This requires a shift in thinking. Decades of living in a technopoly have taught us to feel shame in ever proposing to step back from the cutting edge. But, as in nature, productive evolution here depends as much on subtraction as addition.

It's difficult to understand the influence of technology on your own mind. I think I keep looking for the straw that breaks the camel's back, and I'm strongly suspecting AI is it for me.

How to reverse AI brain rot — A note to myself:

Keep skills alive that a machine could easily replace, even if it's inefficient. Cooking, writing, repairing, designing etc. Automate everything else if you don't see value in it. But don't just over-automate for the sake of it.

Examples include:
— Try to fix things when they break, sewing, wrenching, have fun with it
— Write long form articles just for the sake of it, it helps you think better
— Edit photos yourself instead of using presets, train your eye and mind
— Write code or design by hand even if AI could do it for you. If you find a task that you actually enjoy, do it manually even if it takes longer.

It’s good to see reminders that what we actually need more of are ways to remain human, and practice human skills. I’m increasingly of the mind that sitting at a computer for 8+ hours a day is decidedly not human.

How to usher in an era of abundant donuts • Buttondown:

What would you say to someone who proclaims, “I want to be a donut maker,” but has never actually made a single donut in their life?

You say “That’s awesome. What can I do to help?”

This is a beautiful essay.

Team Mirai and Democracy - Schneier on Security:

Japan’s election last month and the rise of the country’s newest and most innovative political party, Team Mirai, illustrates the viability of a different way to do politics.
In this model, technology is used to make democratic processes stronger, instead of undermining them. It is harnessed to root out corruption, instead of serving as a cash cow for campaign donations.

This is fascinating. I don't know enough about it yet, or have thought through it enough yet to say if it's good, but it is fascinating.

I’m OK being left behind, thanks! – Terence Eden’s Blog:

There are a 16,000 new lives being born every hour. They're all starting with a fairly blank slate. Are you genuinely saying that they'll all be left behind because they didn't learn your technology in utero? No. That's obviously nonsense. It is 100% OK to wait and see if something is actually useful.

Seconded.

How to do the work • Buttondown:

The cruelest thing the tech industry ever did was to tell you that they cared about you. They built you nice campuses, they called you family, they gave you clothes with their name on it. They fed you, they washed your clothes, they got you to ride in their Pride floats. They made you feel like you had not just a job, but a community. And yes, they paid you well. The stupidest thing we ever did—and I say this with nothing but love for you in my heart—but the stupidest thing we ever did was to believe it. IT was neither true, nor never-ending.
The same industry that once called you family is now using the fruits of your labor to commit war crimes. The same industry whose leaders once posted front-page missives to their sites about doing a better job in terms of diversity and inclusion are now selling their technology to fascists who use it to bomb schools.
The industry has decided what it wants to be.

Daring Fireball: Thoughts and Observations on the MacBook Neo:

And the PC world should take note. One of my briefings today included a side-by-side comparison between a MacBook Neo and an HP 14-inch laptop “in the same price category”. It was something like this one, with an Intel Core 5 chip, which costs $550. The HP’s screen sucks (very dim, way lower resolution), the speakers suck, the keyboard sucks, and the trackpad sucks. It’s a thick, heavy, plasticky piece of junk. I didn’t put my nose to it, but I wouldn’t be surprised if it smells bad.

Damn, John… tell me how you really feel!

I think he's right about this new MacBook Neo, Apple is going to sell boatloads of these, and my bet is that they'll get great reviews.

Microsoft's new 10,000-year data storage medium: glass - Ars Technica:

There’s been a lot of preliminary work demonstrating different aspects of a glass-based storage system. But in Wednesday’s issue of Nature, Microsoft Research announced Project Silica, a working demonstration of a system that can read and write data into small slabs of glass with a density of over a Gigabit per cubic millimeter.

I’m not a big fan of Microsoft, but this is just flat out cool.

Owning a $5M data center - comma.ai blog:

These days it seems you need a trillion fake dollars, or lunch with politicians to get your own data center. They may help, but they’re not required. At comma we’ve been running our own data center for years. All of our model training, metrics, and data live in our own data center in our own office. Having your own data center is cool, and in this blog post I will describe how ours works, so you can be inspired to have your own data center too.

Things have changed since 2014 when I was last administering a data center.