# Field Notes No. 4 By Jeff Stowell · 2026-03-26 · Insights > A grand theory of "distribution," some venture data from 2025, my concerns about where we are headed in public health, and some thoughts on AI. # Field Notes No. 4 This week has more AI in it than I would like... not because it's all the rage, but because it has "broken contain" and is impacting all of our daily lives. I made this website with AI! But I do write these missives all on my lonesome. Which is why there are typos. Enjoy! ## **Something for Founders** I have been operating from my own grand theory of distribution for awhile. It goes something like this: the internet didn't open up new channels for reaching customers, it killed all the old ones and replaced them with two. Before the commercial internet, distribution was multi-channel. You gave haircuts? You advertised in the local paper, put a sign out front, maybe ran a spot on the radio - let "Your Fingers Do The Walking." (The Youths™ won't get that one.) You ran an HVAC company? Direct mail, truck signage, word of mouth, local TV, maybe a Pennysaver ad. A Moving Company? AA Moving Company. AAA Moving Company. You opened a burger joint? All of the above, plus whatever foot traffic the location itself bought you. Every business had a patchwork of channels, and savvy business owners knew how to work all of them. Then the internet arrived. And for a while, the patchwork got richer. There were directories and banner ads and email lists and early social platforms. More options, more leverage. It felt like a golden age of distribution. Then Google and Facebook ate everything. Today, if you run a salon or barber shop, you are essentially distributing through Google and Facebook. HVAC company? Google and Facebook. Burger joint? Google, Facebook, and maybe a little Yelp if you're lucky. Software business? Same story, except you're also paying four times as much per click. The entirety of business, every category, now routes through two tollbooths. Multi-channel became captured-channel. That was fine for awhile because at least the roads had rules. Buy ads. Write meaningful content. Attend to keywords. Get good backlinks. You too can "rank!" AI has rapidly changed all of that. Cloudflare CEO Matthew Prince [laid out the numbers at an Axios event in Cannes last year](https://www.axios.com/2025/06/19/ai-search-traffic-publishers), and they're striking. Ten years ago, Google crawled two pages for every visitor it sent you. That ratio had already slipped to six-to-one just six months before his talk. By the time he spoke, it was eighteen-to-one. The AI crawlers are even more extractive: OpenAI's bots were sending one visitor for every 1,500 pages they scraped. Anthropic's were sending one visitor for every 60,000. To paraphrase, people aren't following the footnotes and that's where your traffic has been relegated. Your content is being consumed. Your traffic is not coming back. The tollbooth didn't just get more expensive, the road has been quietly rerouted. This was a year ago... what is it now? Why does this matter to founders? "Build it and they will come" was always a fantasy, but at least there used to be a known way to get customers to notice you. Now, every dollar and hour of effort you spend on customer acquisition is a contribution to someone else's ecosystem. What can you get for that dollar? The founders I see navigating this best are the ones who treat owned distribution like email lists, customer word of mouth referral, community, and events as a strategic asset equal to their product. A necessity. The window to build a direct audience that doesn't route through someone else's tollbooth is sort of open right now. I don't know how long it stays open. _What percentage of your new customers found you because of something you own and control, versus something you're renting from Google or Facebook? for those doing paid acquisition, how do you handle your new friend you bought? If the answer makes you uncomfortable, focus on that._ ## **Something for Investors** If you're an active investor, or thinking about becoming one, [Carta just dropped their Q4 2025 VC Fund Performance report](https://carta.com/data/vc-fund-performance-q4-2025/), and it's worth a few minutes of your time. The headline is cautiously good: median net IRRs for the 2021 and 2022 vintages have climbed out of negative territory, now sitting at 1.4% and 0.7% respectively. A year ago those numbers were negative. So the worst-performing vintage cohort in recent memory is at least trending the right direction. That's something. But what actually jumped out at me: the LP base is shrinking pretty quickly. Back in 2022, the median VC fund between $100M and $250M in assets had raised capital from 83 different LPs. By 2024, that number had fallen to 47. This is happening across all fund sizes. For the 2025 vintage, the median VC fund had just 23 LPs, less than half the median LP count from four years before. Fewer people are actively deploying into VC, and this contraction is likely a symptom of lessening demand for the asset class among some allocators. Illiquidity has a long tail, it seems. When the broader venture market entered a downturn in 2022, many LPs chose to reduce their exposure or pull out entirely. And the capital that IS flowing? It's concentrating. The LPs still participating are writing bigger checks. The median anchor check for funds in the $100M-$250M range rose from $23.1M in 2022 to $35M in 2024. The market isn't disappearing. It's consolidating around fewer players. For those of us investing at the smaller end (seed, pre-seed, syndicates) this is the world we operate in and might be our new normal for awhile. Fewer LPs in the ecosystem means more competition for later financing among fewer funds. DPI remains suppressed across the board, which exacerbates the illiquidity already being felt. I hesitate to use the phrase "circling the drain..." but this is a rough time for the industry. One more number I keep turning over: in the upper-tiers of performance, the smallest venture funds often post the best returns. In the 2018 vintage, the 90th percentile TVPI among funds with $1M-$10M in assets is 4.03x. For funds in the same vintage with $100M or more in assets, the 90th percentile TVPI is just 1.67x. Power law games are easier to win in smaller chunks. I like it when numbers agree with my world view. :) The gap between the top performers and the rest of the field jumps out as well. I can't tell if that's simple math where more capital means more shots at a power-law outcome, or whether there's a genuine management advantage at scale. Probably some of both. _What jumps out at you from this report?_ ## **Something I'm Reading** ProPublica published a piece this week;[_"How RFK Jr.'s Vaccine Agenda Risks a Return of Childhood Plagues"_](https://www.propublica.org/article/rfk-jr-vaccine-agenda-childhood-plagues) . It's frightening on several levels. I used to work in public health. Specifically community-based public health, the part of the system that's not hospitals and doctors' offices, but the un-glamorous machinery that actually prevents disease at a population level. I spent enough time in that world to understand how fragile it is, and how long it takes to build. What is currently happening to our public health infrastructure is alarming. The piece opens with a pediatric infectious disease doctor in a New York City ICU in 2022, staring at a gravely ill infant with a bacterial infection called Hib -- Haemophilus influenzae type b. A vaccine introduced in the 1980s had made Hib so rare that he'd never seen a case in his career. And now they were rolling in. This is a story of failure at scale. A literal unwinding of success. The piece documents how vaccine-preventable diseases have come roaring back in places where access or trust faltered, and draws a straight line to the policy direction we're now heading. I am alarmed. The epidemiological history in here is important and the human stories are hard to overlook. I spent months toward the end of 2025 trying to get some communities I live in and care about to join together to _at least_ replace the disease monitoring that the federal government had so reliably done on our behalf for decades. I was largely met with crickets. I want you to know that this is dangerous. It's dangerous for people and it's dangerous for business and whole economies. This is the kind of story that should make you think differently about what "public health infrastructure" actually means, and what it costs to rebuild once it's gone. I would encourage all of you to at least understand how this is unraveling in your community. ## **Something I'm Thinking About** More AI. I mentioned last issue that I've been spending quite a bit of time trying to refine my approach to the world under my new AI Overlords. I am trying to approach all of this with patience. Lots and lots of patience. Not because I don't think what's being built is important. There are some really remarkable tools being built with an uncommon swiftness. I'm being patient because the noise-to-signal ratio right now is insane, and the people making the loudest claims have the strongest financial interest in you believing them. What I keep coming back to as an investor is that the foundational models, the ones coming out of Anthropic, OpenAI, Google, can have a democratizing effect in a way that feels genuinely new (though there's some interesting new data suggesting these tools are doing more to help the rich get richer... more on that at a later time). When the underlying capability is available to everyone at commodity pricing (at least for now), the old concept of "building a moat" doesn't work the way it used to. As I mentioned above, it's not obvious that anything you create and put out into the world isn't being quietly consumed for training purposes. The question I keep asking is: what, exactly, is a "moat" anymore? I don't have a clean answer yet. But I'm asking the question differently than I was six months ago. It's less "what can AI do?" and more "what does it change about what makes a great business or a great product?" That's a harder question. _How are you approaching the New AI World Order?_ Next week... I've been thinking about startups in Europe. Thanks for reading! /JLS --- Canonical URL: https://hayseedcapital.com/blog/field-notes-4 Site: The Hayseed (https://hayseedcapital.com) — writing, coaching, teaching, and early-stage investing by Jeff Stowell. Operated by Hayseed Capital, LLC, Park City, Utah.