In pursuit of the optimal organization.
Open Orgs is a search for the optimal form of human and agent organization: long-lasting, economically efficient, socially optimal. Three principles. Three axes. No proprietary framework.
Three constraints
Each principle pushes one axis to its extreme. Taken together they describe an organization where every decision has a route in, every role has more than one holder, and every boundary stays open.
All Affected
extreme consultation
Everyone materially affected by a decision has a route into it before it is drafted: not a veto, a route.
No One Is Indispensable
extreme competition
Every role has at least two people who can hold it. Indispensability is a defect, not a distinction.
Open Systems
extreme integration
Every boundary is an interface. A unit that cannot be joined from outside is not open.
What each principle asks you to actually do
A principle you cannot audit is a preference. Each of the three resolves into the practices below: what to put in place, and the move it rules out.
All Affected
The focus is on decision quality, not on speed. A way in for those most affected exists before the decision is drafted, and the decider still decides.
Domain map
Each domain names an owner and a standing consult list, held loosely current rather than exhaustively correct. The list is who the owner designs with, not who the owner has to clear: putting the most affected and the most knowledgeable in one room is what produces the better decision. It runs on cooperative intent, and a party that will not hold that can be dropped by the owner and the rest of the list, at the risk of losing them.
List the parties who cannot push back: the supplier who depends on your volume, the customer inside a renewal, the team the change lands on.
Circulate a finished decision for comment and count the comment window as consultation.
Objection log
Consultation is only testable if the overruled objections survive. Each objection is recorded with who raised it and the reason it did not change the outcome, and the log stays readable to everyone the decision affects. Being able to see what was argued against is what separates a route in from a suggestion box.
Close each decision with what was heard, what was chosen, and the risks the decider still carries.
Retire an objection by waiting the objector out, or by editing it out of the notes.
No abusive terms
Do not hold a party to terms you would refuse yourself. Repeat business, referrals, and the benefit of the doubt when something breaks all come from the same place: the other side leaving fairly treated, every time. A win they would not have agreed to with open eyes is a loan against the relationship.
Price where a party with alternatives would still choose you, and make every clause cut both ways: same notice, same exit, same penalty.
Win on the paperwork: auto-renewal on a window nobody catches, price changes only one side can make, exit fees, payment terms your supplier has to borrow against.
Owned by the affected
Ownership and board seats go to the most affected of each class, elected by it and guaranteed in the constitution rather than in a policy. Associations and public benefit corporations are built to be owned this way. Not every affected party needs to be inside the structure: the most involved are enough to hold a class well.
Name the affected classes, give each an elected seat, and put the guarantee where a later majority cannot vote it away.
Draw the affected classes narrowly enough that the people already in charge are the only ones who qualify.
All Affected guarantees a route in, not a veto. Averaging every position produces a decision nobody argued for, so the elected role still decides, and stands behind it: the alternatives weighed, the reason chosen, the objections not resolved.
No One Is Indispensable
The focus is on tenure, not on people. A role is held for a stated term, and the term is what makes a better holder findable.
Every role has a second
Indispensability is something an organization builds, not a quality a person has. So every critical role is assigned a second: a vice who supports the work week to week and is trusted with calls of their own. Without one, the role cannot be put up for renomination, only defended. Competition for a role starts with there being someone to compete.
Rotate the second holder in for a real cycle, and write the role's knowledge where the next holder could find it.
Name a successor who has never made a decision the role owns.
Fixed mandates
Every role is held for a stated term with a stated end date. Founder is not a role: it is a fact about who started the company, it stays true for life, and no seat comes with it. A founder who is still the best CEO keeps the job by winning it again, a stronger claim on it than never having been asked. What ends is the mandate, not the relationship: there is a handover, and the outgoing holder is placed where their experience is now worth most.
Publish each mandate's start, its end, and the criteria the next review will measure.
Treat "nothing has gone wrong" as the case for renewal.
Open renomination
Before a mandate ends the role is opened, and the incumbent applies for it with a written plan like anyone else. The committee compares plans against published criteria, not candidates against the incumbent. Continuity is a legitimate result. It just has to win.
Decide after the applications are in, not before, and publish which criteria the winning plan met.
Seat the incumbent, or the person who picked them, on the committee that decides the renewal.
Constant rotation has a real cost: everyone who depends on a role has to get used to a new person and new ways of working, and a promising change often underdelivers. So a mandate comes with a minimum stay: the holder cannot apply for another role until it is served. That length is the setting, longer for stability and shorter for change. The competition is continuous. The switch is not.
Open Systems
The focus is on the unit boundary, not on secrecy. What crosses it is decided in advance and written into the contract.
Interface inventory
An interface is a place where a party can contribute without a prior relationship. It documents how to interact with it, names the role that owns it, states who may use it, and sets the terms. That covers code and APIs, and equally roles, teams, contracts, suppliers and subcontracting. Written down, the work belongs to the interface rather than to whoever happens to be doing it, so it can be measured and taken over by anyone who can meet it: a colleague, a supplier, a system. The inventory is the interfaces still unwritten, with an owner and a date against each line.
Document each interface as if for a stranger: how it is used, the role that owns it, who may call on it, and on what terms.
Count an internal interaction as an interface: nothing is an interface until a party outside the unit asks for it.
Right of access
Every affected party, employee, partner, supplier or customer, has a named route to the records that bear on them: accounting, role criteria, subcontractor terms. Secrecy is justified by the harm it prevents, never by the scrutiny it avoids.
Put the request route in the interface terms, and make every refusal name the harm that outweighs the stake of the party asking.
Mark a record confidential once and let the label answer every party who asks after. Answer anywhere but the interface and there is no route, only a person.
Exit test
A partner or customer can leave with their data, their configuration and their history, in a format the next provider can read, and the path out is published and timed. Revenue that survives that test is paid for value delivered. Revenue that only survives the cost of leaving is paid for a moat.
Publish the export, the switching path and how long it takes, then check it by running it.
Engineer incompatibility and call the result a competitive advantage.
The claim runs to affected parties, not to the world. A competitor is not an affected party. A supplier whose payroll depends on you is. So the boundary is planned rather than dissolved: what is open, to whom, and under which contract, is settled before anyone asks.
A constraint on structure, not a fit for a sector
The principles constrain how decisions are routed, how roles are held, and where boundaries close: not what an organization makes or who it answers to. There is no membership tier and no certification. Adopting them means keeping a practice, not buying one.
Two ways in
Adopt the principles and record it, or talk it through first. Neither costs anything, and neither commits you to the other.
Sign the principles
Adopt all three principles and record your organization and the date it adopted them. Partial adoption is not listed.
Request a consultation
A call about what adopting the principles would mean for your organization: where they would bind, and what would have to change. No cost, no engagement.