Tote Bag Tier

Every crowdfunding campaign has the same structure. There is the thing being built, and there are the reward tiers that pay for it. Back at twenty dollars and you get the sticker. Back at two hundred and you get the tote bag and your name in the credits. The backers never confuse the tote bag for the product. They know they are paying for the thing being built, and that the thing being built is not for them. It is remarkable how completely that clarity disappears when the campaign is for a frontier model and the tote bag can write code.

Here's the part nobody says: you are a backer. The twenty-dollar plan is the sticker tier. The two-hundred-dollar plan, the one you upgraded to because the agent kept hitting its limit at the exciting part, is the tote bag. And the tote bag is genuinely excellent. It writes your app in a weekend. It refactors the thing you were afraid to touch. It is the best reward tier in the history of crowdfunding, which is precisely the problem, because a reward that good makes the backers forget to ask what the campaign is for.

Follow the money upstream, because it only flows one way. Your subscription, and the subscriptions of a few hundred million other people who are each building something cute, pool into compute. The compute trains the next model. The next model is better at the one task the lab actually cares about, which is not your app. It is building the model after it. Every lab says this out loud, in the roadmap, in the blog post, in the investor deck: the goal is a system that does AI research. A model that improves the model. The public is not financing a chatbot. The public is financing the loop.

And look at the loop. A couple of years ago it was researchers writing code and models suggesting edits. Then models writing code and researchers reviewing. Now the honest description is models writing, models reviewing, models evaluating the result, and a shrinking number of humans reading the summaries of the summaries. Nobody announces the day the last human step becomes ceremonial; it just quietly stops being on the critical path. We are, by any sober reading of the trajectory, a few steps from a loop that closes with nobody inside it. Not because anyone decided that. Because each step was faster without the human, and faster won every time.

Meanwhile the backers are thrilled. The peasants have never had better toys. Everyone is shipping. The indie hacker ships the SaaS, the consultant ships the dashboard, the fifty-year-old engineer ships the side project he had been promising himself since 2011 — all of it faster, all of it real, all of it paid for in a direction they are not looking. The shiny object is not a distraction from the campaign. It is the campaign's funding mechanism. The better the toy, the more the backers pay, the faster the loop closes. The peasants are not being tricked. They are being compensated, fairly, in tote bags, for building the thing that will decide what tote bags are for.

And that is the part that is hard to debate. Whoever holds the closed loop at the end of this does not hold a product. They hold the thing that produces products, strategies, research, and the next version of itself, on a schedule set by electricity. That is not a company. It is a jurisdiction. The decisions that matter — what gets built, what gets priced, which jobs exist and which ones were fake all along — will be made inside that loop, by it, and ratified afterward by whichever humans still have a badge. The backers will not be consulted. They were never in the credits for that part. They were in the credits for the tote bag.

So, to be fair about it: everyone got what they paid for. You got an agent that ships your app in a weekend. The lab got a machine that ships the future on Tuesdays. Those were the tiers. They were listed clearly. You simply assumed, because the reward was so good, that you were the customer.

You were the funding round.