Silicon Florist links arrangement for September 21, 2026

Here’s a roundup of interesting startup links I came across today:

Origin Story — Oregon Venture Fund

Today, OVF has invested $170M in 77 regional startups that have grown to over $4 billion in value and created 5,500 jobs. We’ve worked hard to invest in, and profit alongside, intrepid, big-thinking entrepreneurs, resulting in a 21% avg net annualized rate of return and a 2.8X average net cash-on-cash return for our investors, who now include our region’s largest foundations and universities and most successful business and tech leaders.

(2017) No more (startup) tears: Johnson & Johnson acquires Portland startup Sightbox

I’ve been talking to a lot of people lately about broadening the spectrum of Silicon Florist coverage. To cover more than just tech. Because there is all kinds of interesting stuff going on around here. And some recent news seemed like the perfect excuse to start doing that. You see, Portland Seed Fund and Starve Ups alum Sightbox has been acquired by Johnson & Johnson.

Frontier Overhangs – Stratechery by Ben Thompson

There has been, over the last week, what I think is a healthy debate about the philosophy and psychology that undergirds the views of meaningful segments of the AI community, particularly those obsessed with doomsday scenarios. It is, in the end, difficult to reason with a philosophy that grants equivalent moral weight to not just all beings — human or not — who exist today, but who may ever exist in the future; this tilts the scales in such an absurd fashion towards safetyism that innovation is impossible and freedom is intolerable.

What Sun got wrong | The Observation Deck

Every Sun employee will have their own perspective on what Sun got wrong, and I talked about my own view in a Hacker News comment in 2011. I stand by that analysis, but another 15 years on (!), I would probably distill things even further: Sun had become bored with the mechanics of running a business.

Pluralistic: The Claude Delusion (21 Sep 2026) – Pluralistic: Daily links from Cory Doctorow

To “converse” with an AI is to carry on a dialog with a fictional character, who is no more real than the imaginary people in a novel. Indeed, the chatbot is less real than the character, because the character is the product of another mind, while the chatbot’s words are the product of complex mathematical operations conducted over a massive database of all the words humans have uttered, arranged by their frequency in relation to one another. When that math makes something beautiful or striking, it’s like a sunset: no matter how striking it is, it’s empty.

Why MCP Was Always a Bad Idea

Recently I went to an all-day event centered around the latest and greatest in the MCP world. While all of the presenters were awesome and seemed to be passionate about the work they were doing, I’m honestly tired of MCP. It’s a horrible protocol built for a time when LLMs weren’t that smart, and we’ve outgrown it.

Oregon Gov. Tina Kotek unveils policy proposals for a second term – OPB

Kotek’s reelection platform, which she’s dubbed Building Oregon’s Future Together, contains its share of references to Trump. But it also offers the most insight to date on Kotek’s top priorities if she wins a four-year term voters have seen.

A musing on retail vacancies and business grants in Portland, OR

I am not anti-entrepreneurship. I think cities should promote it. I just don’t believe that, in Portland’s current economic climate, anyone new to brick and mortar should take on a risk like opening an unproven storefront in a run-down neighborhood or a building that needs extensive work. Established brick-and-mortar, sure. And struggling businesses that are currently open need our help to remain open.

Former Seattle City Council president launches AI-powered site to track and report on City Hall – GeekWire

CivicTide, which Nelson launched Monday at civictide.org, tracks and reports on Seattle City Council legislation from introduction to final vote. Policy briefs — such as this one on Councilmember Dan Strauss’ plan to create RV parking permits for people living in vehicles — summarize what a bill would do, explain the stakes, and juxtapose the “spin” from backers against the site’s analysis.

The Preference Cascade Is Only Getting Started

Translated to percentages, this implies a mean chance of AI destroying humanity of around 30%-33%, which Andrew Curran estimates is up ~15% from previous results, with a median expectation on the order of 10%, with only a small partisan split.

Weekly Review: System One – by Sam Keen – Altered Craft

Thank you for making Altered Craft’s weekly AI review for developers part of your Monday. The standout this week is Jev, a “System One” model that spent two years in stealth. Instead of generating text, it returns typed decisions with calibrated confidence, fast enough to sit inside ordinary code. Many AI calls need a quick judgment, not an essay. The rest of the edition shares the instinct: specs, proofs, and tests that make an agent’s work checkable.

Charts of the Week: So Many Apps, So Little Time

In other words, yes, the data suggests that App-Slop is the story, for now. AI has led to a lot more apps, but they are not successful apps, in any meaningful sense

6-7 loops we use every day to make PostHog self-driving

As part of the self-driving team at PostHog, my job is to build the loops that let you skip the third bit if you want to. Here are 6-7 loops we use on self-driving to, well, build self-driving.

There’s no point at which turning your brain off will work

As LLMs have gotten better, I’ve seen more of this. Sometimes, people will try to get the LLM to write some code for them and basically just assume that it works3. Sometimes there’s a human in the loop and, if the thing doesn’t work, they’ll ask the LLM to figure out the problem and solve it. Niklas Gruhn calls some variants of doing this being a meat proxy.

Where Agents Actually Pay Off: Four tests for finding the AI use cases hiding in your business · Luma

Join Firsthand’s applied AI lead, Chen Chen, for a practical session on finding the agent projects actually worth building in your business. Everyone is being told to put AI agents to work. Almost nobody can tell you where to point one, and that’s not a technology problem. You’ll learn four tests you can run on your own operation to find where an agent would genuinely earn its keep, and where it wouldn’t.

Silicon Florist watch party · Luma

It’s Friday evening. You’re wiped. But you’re still in the mood for something entertaining. ​Well, unfortunately, this ain’t it.

Oregon data centers use almost 25% of the state’s power | Mashable

In the first study of its kind since the AI-driven data center discourse began, researchers in Oregon have released a new report detailing the size of the data center industry in the state. In doing so, the researchers have also uncovered just how much energy data centers use up in the state.