Why ADHD Founders Start Everything and Finish Nothing — and the 4-Slot System That Breaks the Pattern

ADHD brains have lower baseline dopamine availability in reward-processing regions, including the striatum and prefrontal cortex (this is consistent across decades of imaging studies — see Volkow et al., JAMA 2009, [PMC2696794](https://pubmed.ncbi.nlm.nih.gov/19738093/)). Lower baseline means the brain is hungrier for spikes. And novelty is one of the most reliable triggers of a dopamine spike — it works even when the novel thing has no real-world reward attached.

Translation: starting a new business idea releases dopamine on day one regardless of whether the idea is good, profitable, or compatible with your existing projects. Your brain isn't evaluating ROI — it's chasing the chemical. Finishing the existing project would produce a much larger long-term reward, but the brain discounts future rewards steeply (Barkley's work on delayed-discounting in ADHD is the canonical reference, [PubMed 1842301](https://pubmed.ncbi.nlm.nih.gov/9000892/)). The result: 'finish the half-done thing' loses to 'start the shiny new thing' every time the choice is offered.

The trick is to stop offering the choice.

You are allowed four open project slots. Not five. Not 'four plus a side experiment.' Four. A slot is occupied the moment you've spent more than two hours on something or told anyone you're working on it. To start a fifth project, you must first either ship one of the existing four to a defined launch state, or formally kill one (delete the repo, archive the doc, move the brand kit folder to /killed/). There is no 'parking' a project. Parking turns into a graveyard. Either it ships or it dies. The slots make the trade-off visceral: you cannot just add — you must choose what to lose.

Why four? Two is too tight for real-world founder reality (you usually have one paying client, one product you're building, one marketing channel running, plus the next idea). Five is too many — empirically, every founder I've watched try five ends up with five half-done things by quarter's end. Four is the largest number where the constraint still bites. It hurts. That's the point.

Every slot has a written launch state defined the moment the slot opens. Examples: 'shipped' for a course = first paying customer, refund issued if asked. 'Shipped' for a SaaS = 10 active users on the paid tier, not 10 trial signups. 'Shipped' for a blog = first 50 organic visits from a non-paid source. The definition matters because 'I worked on it really hard' is not a shipping state — it's a cope. ADHD brains will renegotiate the goalpost in real time if you let them.

Write the launch state in the slot file BEFORE you do any work. If you can't write it, the project isn't ready for a slot — it's still a vague want, and vague wants belong in the someday list, not the slot board. The someday list is unlimited. The slot board is four.

Slot stays open if: the work is on a clear path to the written launch state, and you've touched it in the last 7 days. Movement, not perfection.
Slot must be killed if: you haven't touched it in 14 days OR you can't honestly answer 'what's the next concrete action toward the launch state?' Kill it. Open the slot for something you actually care about.

Every morning, you look at the slot board and ask one question per slot: 'what's the smallest next move toward the launch state for this project, today?' You don't pick what to work on by interest. You pick by what's closest to shipping. Slots near the launch state get priority over slots that just started, because closing a slot opens room for the brain to spike on the next new thing legally.

The ritual is roughly 3 minutes. Open the board, write one next-action per active slot, pick two to actually execute today (not four — your real capacity is two on a normal day, three on a great one). The rest sit. This is also where you decide if a slot needs to die. Two consecutive mornings of writing the same next-action and not executing it = the slot is dead. Kill it that morning and free the dopamine.

**The 'this is technically the same project' trick.** Your brain will try to argue that the new course is really a different module of the existing course, so it doesn't need its own slot. If it has its own landing page, its own promised outcome, or its own launch state, it's a new slot. No exceptions.

**Client work isn't a slot — it's the foundation.** Slots are for things YOU initiate. The retainer that pays the rent is below the slot board, not on it. Otherwise you'll claim three of your four slots are 'client work' and never build anything of your own.

**Hyperfocus binges feel like cheating the system.** They aren't. If you finish a project in three days of hyperfocus, you've earned the open slot. The system only breaks when hyperfocus produces a half-done fourth thing while three other slots also sit half-done.

**'Research' as a slot is a trap.** 'Research the AI agent market' is not a project. 'Ship a public scorecard of 20 AI agent tools by end of month' is. If you can't define the launch state, it's not a slot — it's a Google search.

**Killing slots feels worse than starting them.** It does. Kill them anyway. The mental cost of carrying three zombie projects in your peripheral vision is higher than the one-time grief of formally killing them, and every dead slot you avoid killing is a slot you can't use for something alive.