Building on this post, I like the point made at the end of the episode.

The experiences are different because Sam’s Club zagged to Costco’s zig.

It makes sense to not replicate what your chief competitor is doing but offer something different and distinct because you’ll appeal to a different segment of consumers.

Find value for your consumers that’s not just being the cheapest product available, because you’re never going to win that race

Give consumers are compelling reason to shop at your store and trust your brand

Do the same thing as everyone else and get worse results.

we didn’t ban noisy aircraft we banned supersonic aircraft

Incentives matter and words matter.

Here, poor word choice disincentivized an entire industry.

Clarity of messaging helps align incentives.

via Alex Tabarrok

TV is now the primary device for YouTube viewing in the US.

and according to Nielsen, YouTube has been #1 in streaming watch time in the U.S. for two years

And Mr. Beast has more subscribers than Netflix.

YouTube is mass media via niche media.

via YouTube

As Seth said:

Discovery doesn’t just happen in the online store’s search box.

I would go further and say true discovery never happens in the search box.

Awareness and exposure—the precursors to intent—happen elsewhere. In the rhythms and routines of life separate from shopping.

Anthropic released the Anthropic Economic Index

The Anthropic Economic Index aims to understand AI’s effects on the labor market and broader economy over time. The Index provides the clearest picture yet of how AI is being incorporated into real-world tasks across the modern economy.

Trends are based on anonymized usage data for their Claude model.

Seems like a useful indicator to add to the monitoring mix.

All advertising is unwanted, so if you’re going to crash the party, bring some champagne.

-Bob Thacker

DeepSeek was just the beginning 🤖

AI researchers at Stanford and the University of Washington were able to train an AI “reasoning” model for under $50 in cloud compute credits

Training methods appear to be proliferating with the number of models available:

The s1 paper suggests that reasoning models can be distilled with a relatively small dataset using a process called supervised fine-tuning (SFT), in which an AI model is explicitly instructed to mimic certain behaviors in a dataset.

A race to the bottom we can actually benefit from:

After training s1, which took less than 30 minutes using 16 Nvidia H100 GPUs, s1 achieved strong performance on certain AI benchmarks

& a lesson we can all learn from:

The researchers used a nifty trick to get s1 to double-check its work and extend its “thinking” time: They told it to wait. Adding the word “wait” during s1’s reasoning helped the model arrive at slightly more accurate answers

via TechCrunch

Chesterton’s Fence

There exists in such a case a certain institution or law; let us say, for the sake of simplicity, a fence or gate erected across a road. The more modern type of reformer goes gaily up to it and says, “I don’t see the use of this; let us clear it away.” To which the more intelligent type of reformer will do well to answer: “If you don’t see the use of it, I certainly won’t let you clear it away. Go away and think. Then, when you can come back and tell me that you do see the use of it, I may allow you to destroy it.”

Or, as Nudie says:

Not all change is progress