![[Matter Header Twitter.png]] Publish Date: [[Wed Aug 19 2026]] [Here's a trip down memory lane](https://podcasts.apple.com/us/podcast/the-gray-area-with-sean-illing/id1081584611?i=1000779239467) with Sean Illing for those interested in where 21st century media (the Internet) has been, and where it might go. Also deeply instructive for practitioners. #### How ads got us to how we use a (desktop) computer today *"Things on the Internet had to be seen, or sent to you - like an analog"* That's the before times. Anyone - most of you - who remember loading a floppy disk, or hearing the crazckzzizzz of a telephonic Internet connection probably do so with some degree of fondness.  There was a physical nature to using the old Internet that has been lost to advancing technology and design trends. Old computers - keyboards - felt like tools. You might take them apart with a screwdriver and put memory chips in yourself. And it felt more like work to discover things, even if that discovery was entirely for play. Google took the *effort* out of said discovery, in exchange for your attention. Which is arguably the single core innovation 26 years into the 21st century Internet; and created many trillions of dollars of wealth. Facebook took the effort out of spying and gossiping in your social circle, in exchange for your attention, and billions in valuation. Spotify took the effort out of finding music, in exchange for your attention; and made vinyls niche for hipsters. *Et cetera, ad nauseam* Now, this paradigm of Internet usage is melting into a mess of [[On Political Botslop|slop]]. #### The Internet’s "Frame" Back to the podcast: Broderick's point about recipe blogs draws the power of Google’s revolution out: people write a certain way about recipes because Google *wants* them to write a certain way. It was necessary. Google's dominance at surfacing information broke the form of conversational discussion about cooking, to ensure people found a particular author's recipes. If it's hard to find recipes online, it's a lot harder to sell ads for cookware. Google made [[Marshall McCluhan]]'s "Medium is the Message" (1964) fully manifest: when we create for the purpose of surfacing information; the machine becomes the audience. When Google essentially made the whole of the world wide web a commercially available platform, the advertising needs of the platform, instead of the technology itself, dictated the form of the media displayed. LLM's are taking that insight a step further, to a level of mechanical correctness. It doesn't take long today to find entire websites created by AI on esoteric topics, designed to pull traffic and hawk ad traffic at near-zero marginal costs. The machine creates the audience to mimic the audience using the machine. The user is a third party to the whole dynamic. So the user experience became steadily messier. Until you get this: ![[I'm Feeling Lucky.jpg|456]] Remember Google's early feature "I'm feeling lucky;" which would just take you to the top web result in your search? That's certainly a low friction application of search technology. It's just a lot harder to sell ads against. So it's long gone. ## Does the 'Death' of Search really mean the Death of the Browser? Back to the Podcast: the Internet of the 2010's as "a black hole for millennial labor" rings true to someone who was just shoveling political budgets to Facebook and Display ads for much of the decade. I ran millions of dollars of them for campaigns around the country, winning and losing. It was standard practice. *People were on computers and iPhones at least as much as TV; so let’s run ads on computers and iPhones however we can.* But it was met with a nagging suspicion: did video ads in the browser or the first decade of mobile, static ads ever really work, especially for political persuasion? Sure, click here to order a pizza or click here to... buy cookware. Those worked for a bit. Sometimes still do, for point of sale marketing. Not so much for political persuasion. Contextual display style ads on the Internet, even with motion, just don't work as well as TV. They never did. Our eyes in a browser are conditioned to moving fast, always further down, scanning. On mobile, scrolling. It's never been akin to sharing a physical space with billboards or holding a newspaper; where your eye will scan and then fix, moving with your body. No computing based platform ever showed up in verbatim polling statements or moved any numbers on a ballot test, until YouTube came along. Even then it was limited, but you could target narrow (GOP primary audiences) with unique messages, and have them spit back to you in open testing. Which makes sense: YouTube focused the viewing environment for someone *who was actually ready to watch a video*. Which lasted about four years as a “computing” phenomenon. Youtube, of course, just [[Everything Is Television|became TV]]. The eye has to hunt for the thing it actually wants - so by the nature of our attention, we find a way to quickly sort out and exclude everything else. Flashing ads first among them. That world - that economy - is slipping away. Firstly, talk about a loss of physical culture: [AI chat might kill the keyboard](https://www.thetimes.com/business/technology/article/ai-voice-text-death-keyboard-gzwdmdl86) (...cue a cold dead hands pun). But more broadly: think of your user experience on the open Internet today. Per above; it's a mess. ### We talk about politics the way Google wants us to talk about politics, whether we like it or not. That sounds more revelatory than it is: Google wants us to talk about politics through the lens of the media institutions which have been filtering politics to us for the last hundred or so years. Search, up to now, has been a *centralizing* force in our politics. It made the local national. It needed to rank things, so it ranked the things according to information gatekeeping and brand cachet. Your eye would get caught by names and logos you knew - so Google placed them right below one or two lines of ads. WSJ. Washington Post. Fox News. NBC. The Guardian. These are not internet-first institutions. But when you Google a political topic, their legacy web properties pop up top. When I Google the Iran war with location services on from my home in Indiana, this is what I get: ![[New in the Iran War AI.png|381]] Yes, Google News has made some oblique nods to including local content sources. But, for the most part, if you search something political \- unless it is *truly* local \- you'll get a national news outlet's take on it. Even if you might be interested in a local angle. AI is changing that. If I *specify* Indiana in a similar query, this is what I get: ![[Iran War Affecting IN AI.png|451]] Here's the rub: they key word there isn't "Indiana." It's "affecting." When I apply a gerund to my question, Google's LLM "knows" I want "context," so it spits some out. That's actually a back to the future move on political information flow. Until about twenty years ago, everything I knew about the Iran war was filtered through Hoosier perspectives, Hoosier media, Hoosier related stories. If I wanted the Washington Post, it was available but not one click away. AI is making the local possible again. It might just restore the promise of the early Internet - able to surface things in nooks and crannies otherwise. Or it might push us farther into individualized information niches. We bet more heavily on the former: the platforms returning the AI emboldened search queries, Google or not, will need a [return on all that investment](wsj.com). The possible upsides for advertisers should be plain, and are coming: "Here's what Todd Young or Jim Banks says about the Iran war," Mr. Indiana Voter. *Clip Paid for by IN GOP, rendered in-stream as pre-roll to custom search results.* #### Changing the Point of Sale A user poses a query \>\>\> the platform provides the answer the advertiser wants. It’s been working to sell cookware for decades now. It's just never been quite there for politics, outside of recruiting donors… who, much like those perpetually burning their crostini, are online ready to take action. So they need to search to find the donate link. Much simpler than changing a vote preference. Largely because people don't sort out their political preferences like they shop for cookware or look up recipes, point-of-sale *persuasion* in politics isn't much of a thing. We haven’t had the mechanisms, beyond a few lines of search ads and a hyperlink. Especially at the local level; [[Claude, for whom do I vote?|personalized political information flow]] will 1. Become a thing to be managed by practitioners/advertisers and 2. function like a series of DMs from bots, delivering information summaries unique to a person’s geography, preferences, and user history. 3. have us talking a lot about [[contextual persuasion]]. Which isn't about digital targeting. Advertisers have been finding ways to meet their market at point of sale since since the first concessioner wrote "*Vinium - 2 Denari " in chalk above the urinal in the Roman Colosseum. We just haven't been able to package up creative for a meaningful corpus of persuadable/actionable voters and advertise to them like we're sending a DM. It's coming. How do we, ahem, slide in? Other than making shorter form TV, that's the core question for us in the next three cycles; whether primary or general; persuasion or turnout play: how to make enough timely, compelling and *[[Humans Learn Visually|human]]* ads to fill those nooks and crannies. And thus take advantage of all sorts of opportunities inevitably opened up by the commercial markets. Political platforms almost always follow mature commercial innovations. Remember field work before ubiquitous smart phones? No Stripe; no Win Red. Prediction: billions of dollars flow through our tech-media economies in presidential years. Adapting these new tools to capture as much value will line up with the 2028 race, and we'll end up with bigger, more durable answers to these questions.