Some years ago I invested in a friend's business and quickly the annual dividend grew, long since repaying the initial investment. When I get the dividend check each year I like to look at my 2003 tax return to remind myself how lucky I am to be where I am today.

Nike's troubles are quite simple and of its own making. A good mantra in retail is don't make it hard for me to give you money. Nike makes it nearly impossible. A few years ago it retreated from most online retailers and many brick and mortar to become mostly D2C. Remaining third-party retailers could offer only part of the catalog, retaining specialized products for Nike direct. The third parties had wide Nike selections online and in-store, ecommerce had great filters, people could find what they want, or at least get close. Discoverability in ecom is hard. We had 6000+ skus at Zappos. We constantly fretted over colors (display all in search or just on product page), words, category definition, searchfilter facets. What language to customers use to describe what they are looking for? What is a white shoe vs a shoe *with* white? Nike doesn't even try. The site has 779 mens shoe choices. It's impossibly difficult to find the filter you need. Every colorway is displayed as unique and almost never adjacent to its other colorways. Its stores stock a comical, seemingly random array of products. To effectively shop at Nike you need to hear about a shoe from someone else. It blocks ChatGPT and other bots so even if you're looking for something new (find me a casual shoe), ChatGPT can't direct you to Nike. I get this for third-party sites...minimize comparison shopping...but for brands it makes no sense. In store, for example, you can't buy a Nike soccer or tennis shoe at either of its Austin stores right now. That's an edge case but if its a category they care about, it would seem they would use their real estate to at least show you it exists. Try buying a casual Nike at Nike Soho, the flagship store. Nike stores are not a place you go to get Nikes when your intent is utility. They are hype beast outlets designed to brand build on top of very expensive real estate (Nike Austin is next door to Hermes). Brand building makes sense when your product is easy to buy. If you make it hard, you're toast (or Hermes in which case you won at life).

when i'm a billionaire my act of public service will be to run around the country and fund lawsuits like this

Longshoremen’s Union: If you automate anything we will go on strike and cripple the US economy. China:

Just because something doesn't exist and could exist doesn't mean it should exist.

The Austin airport Uber pickup is bad. But you have to understand that Austin was one of the last cities to allow Uber. In the 2010s when Uber was everywhere else, Austin had ~800 taxi licenses. 800! SXSW annual attendance was 432,500 at the time. So it could be worse.

My car has better weather radar than my iPhone.

God blessed Texas

In less than 24 hours of use ChatGPT Dot is my favorite thing OpenAI has ever built. At first I was like why do I need another chat that just layers on top of chats, but I tried it anyway. What's most interesting is what it chooses to surface and what it doesn't surface. 70%+ of my ChatGPT app usage is pushing code to three full-stack platforms that dozens of people use every day, including a robust fleet of agents for my team. It hasn't mentioned any of that. Instead it's looked at my calendar and todo list, but even doing that, its prioritization choices are fascinating. It saw a voice memo my ops agent dropped into todoist about the land and tax structure behind a huge resort development that I wanted to research for a resort-like project we're working on a proposal for. It linked that memo to the proposal and LOI and offered to research and integrate the findings into our pitch. That required seeing a todoist item, a few granola meetings, some Drive docs. It spun up ChatGPT Work sessions to research and later update the LOI, the pitch, and our internal notes. The other example is even cooler. This morning while I was at the gym it sent an analysis of a feasibility study for a review call scheduled this afternoon. It saw my calendar, cross referenced the invite to an email with the study attached, then found the internal notes and a granola meeting involving the JV partner (who is not on today's meeting invite or the email with the feasibility study). These two examples are just 18 hours in. I have a few dedicated agents in my own Hermes fleet with proactive responsibilities and none of them come close to this. Excited to see where Dot goes.

Today I had to push 226 lines of code so my agent could access a payment form that was buried two iframes deep. So this is an improvement.

These are the kinds of compatibility issues that stand between agents and autonomous productivity. But honestly iframes in iframes in iframes is just bad overall, not just for agents.

My Austin city council district election is a choice between: - a guy who has only ever worked for environmental non-profits - a guy who has only ever worked as a local political staffer - a woman who initially filed to run in the wrong year - a woman who forgot she's running

Imagine being a Waymo, with all of your protrusions and cameras and such, pulling up to an intersection next to a sleek Cybercab with no steering wheel. I would be so embarassed.

First time in a car with no steering wheel and the main thing my mind honed in on was the lack of rear view mirrors when it changed lanes.

Am I just so online that I can immediately spot this as AI slop or does everyone else see it too?

Post-workout, post-lunch nap is GOATed. Perhaps second only to the post-long-run-couch-half-nap.