You have asked four different investors for introductions into the same small circle. Every ask was reasonable. Every one of them landed. Then in month six the replies start coming slower, and two of them stop coming at all.
Nothing about you got worse. You were drawing on a shared resource, and you were never the only one drawing.
Why these three models
The decision is how hard to draw on something other people also draw on. A small investor community. A pool of forty engineers who can actually do the thing. A mailing list. One senior person’s attention. The features that fire are a resource that shrinks when used and cannot easily be fenced off, a standing inside it that fades unless it is refreshed, and the occasional draw large enough to end your access in a single move.
Three lenses, and they point in different directions on purpose. The commons produces a complex answer about why the pool fails without anyone behaving badly. Reputation decay produces a cycle answer about why your position inside it is a flow rather than a balance. Ruin sizing produces a random answer about why the average ask is the wrong thing to think about. The second and third are where founders lose access, because both feel like relationship management and both are actually arithmetic.
1. The resource nobody owns is the one you use most
A commons has two properties at once. Use by one person leaves less for the next, and keeping people out is difficult or costly. That combination is what makes it fragile, and it is a much narrower category than it first sounds.
Test it before you assume it. An article you publish is not a commons; my reading it does not reduce your supply. A specific investor’s willingness to make introductions is a commons; every introduction spends credibility they cannot get back cheaply, and they cannot easily refuse the next founder without cost. So is the attention of the three people in your market who answer cold emails. So is the goodwill of the one distributor who will stock an unproven brand.
The same structure sits in places that do not look like networks at all. One distributor who will stock an unproven brand in a market of four is a pool. So is the single procurement officer in a ministry who takes calls from small suppliers. Both are drawn on by every founder in the sector, both have a rate they can absorb, and neither will ever announce what it is. The tell is identical in each case: the reply that used to come the same day starts coming on Friday.
The familiar version of this story ends in collapse, and that ending is wrong often enough to be dangerous. The empirical record on real shared resources is that many of them are governed successfully for generations, without privatising them and without a state stepping in.1 They survive when specific structural features are present: clear boundaries around who may draw, rules matched to local conditions, monitoring by the users themselves, sanctions that escalate rather than jumping to expulsion, and cheap ways to resolve conflict.2
The laboratory version of the same resource confirms the shape. Groups given a shared pool neither cooperate fully nor strip it bare, and what moves their behaviour most is whether they can see each other’s draws and impose a cost on the heavy users.4
Read that list as a founder and the practical instruction is not be considerate. It is find out what the rules already are. Every functioning network has boundaries and graduated sanctions whether or not anyone has written them down. The slower replies in month six were a graduated sanction. They were the warning stage, and it arrived before expulsion precisely because the system was working.
The thing to do with this is uncomfortable and cheap. Ask the person you draw on what the rate is. How many introductions a quarter is normal for someone at your stage. What makes them stop. Nobody asks, and the answer is almost always given freely, because the rules exist and their existence is not a secret.
2. Your standing inside it is a flow, not a balance
Founders treat reputation as a stock. You do good work, it accumulates, it sits there and can be spent later. The formal treatment of reputation says something less comfortable: when what you do is observed imperfectly, a reputation built on beliefs about your type is not permanent. Over time the audience accumulates observations, the beliefs converge on what you actually are, and the reputation effect wears off.3
The mechanism matters more than the theorem. Reputation persists while the other side is still uncertain about you. Every clean observation reduces that uncertainty in one direction and every messy one reduces it in the other, and the messy ones are not always your fault. Noise in what people observe is enough on its own to erode a good reputation, because noise is what keeps the belief moving.
Which means standing in a network behaves like a flow. Asks draw it down. Delivered outcomes top it up. Silence is not neutral, because while you are silent the top-up rate is zero and the decay continues.
Founders track the draws and never track the top-ups. Count them for the last two quarters. Every introduction you requested is a draw. Every time you reported back what happened, closed a loop, sent something useful unprompted, or made an introduction in the other direction, that is a top-up. Most fundraising founders find a ratio worse than four to one, and most are surprised, because each individual ask was small and each individual silence felt like not bothering anyone.
Top-ups have to be legible to count, which rules out most of what founders believe they are already doing. A generous mention to a third party is a top-up only if it reaches the person. A two line note after an introduction is legible. So is a specific piece of market information they cannot get elsewhere, and so is an introduction made in the other direction, which is the only top-up that adds to the pool rather than to your balance inside it.
And when there is nothing good to report, report that. A founder who goes quiet because the news is bad has confused the top-up with the outcome. The report is the top-up. Two lines saying the introduction went nowhere and here is why costs nothing and is the version people remember you for.
The correction is not to ask less. It is to close every loop you opened, including the ones that went nowhere. An introduction that failed and gets a two line report is a top-up. An introduction that succeeded and gets no report is a draw that never got repaid.
3. One badly sized ask can end the access permanently
The first two models are about rates. This one is about a single event, and it does not obey the same logic.
Expected value reasoning says make the ask if the upside times its probability beats the cost. That reasoning is correct only when you survive to repeat it. When one outcome removes you from the game, the average stops being the relevant quantity and the size of the worst case starts being the only one.
Applied here: the relationship is the asset, and each ask is a bet placed against it. Most asks are small enough that a refusal costs nothing. Some are not. Asking a semi-warm contact to personally vouch for you to their largest limited partner is an ask whose refusal is awkward enough that the contact will quietly stop engaging rather than say no twice. The ask did not fail. The account closed.
Size every request by what its refusal costs, not by what its acceptance is worth.
The operational form is a single line you can apply without judgement. If a no to this request would make the next ordinary conversation awkward, the request is too large for the current standing, and the correct move is to make it smaller or to build standing first. This is not timidity. It is preserving the thing that generates all future asks.
Shrinking an ask is a specific manoeuvre rather than a politer version of the same request. Replace the vouch with a question. Instead of asking somebody to recommend you to their largest investor, ask whether that investor is looking at your category this year, and who they would speak to first. That is refusable at no cost, it produces most of the information you wanted, and it frequently produces the introduction anyway, offered rather than requested, which is the version that carries weight.
There is a real exception and it should be named. When the pool is closing anyway, when the round is ending or the person is leaving the market or you are otherwise finished, the calculation flips and a large draw is correct. Bold draws belong to positions that are already lost, not to positions that are merely slow.
What the three say together
Run them in order and they produce a sequence rather than an opinion.
- Is this resource actually subtractable. If your using it does not reduce anyone else’s supply, none of this applies and you should draw freely.
- What are the existing rules. Rate, boundary, and what triggers a sanction. Ask directly.
- What is my draw to top-up ratio over the last two quarters. Count it rather than estimating it.
- Would a refusal of this specific ask cost me the relationship. If yes, shrink the ask.
Three of those four are answerable this week with no new information from anyone except a single question asked out loud.
Where they disagree
The commons model and the ruin model conflict, and the conflict is not cosmetic.
The commons view says restrain your draw, coordinate with other users, and protect the regeneration rate. The ruin view says that when you are close enough to failure, restraint is what kills you, and a single large draw is the correct play. Both are right inside their own conditions and the conditions are opposite. One assumes you will be here next year. The other assumes you might not.
The resolution is a question about your own position, not about the network. If you have more than two quarters of runway, the commons view governs and the large ask is a mistake dressed as courage. If you have less than one, the ruin view governs and careful rationing is a slow version of the same failure. In the middle, and most founders are in the middle, the honest answer is that the models do not agree and you are making a judgement call rather than a calculation. Say so out loud rather than pretending the model chose.
What none of them contain
All three treat the pool as fixed in size. Real networks are not. Some of the people you draw on actively expand the resource, by bringing new investors in, by vouching for the market rather than for you, by growing the thing they are the gatekeeper to. Drawing on an expanding pool is a different situation and none of these three models will tell you that you are in one.
The signal is worth watching for anyway. If the person you keep asking has introduced two new participants into the circle this year, the pool is growing and your draw is a smaller share of it than the arithmetic suggests. If the circle has been the same fourteen names since you met them, it is not.
Second gap: none of these models handles the case where the resource is a person’s emotional bandwidth rather than their social capital. Reputation decay and commons logic both assume a resource that responds to structure. Exhaustion does not, and the correct response to it is not a better ratio.
The one action that survives the ignorance: before your next introduction request, count your draws and your top-ups across the last two quarters, and close the oldest open loop first. Not the most promising one. The oldest one, because it is the one that has been decaying the longest and the report is owed regardless of the outcome.
Who has to move
This is the founder’s own job and it cannot be delegated to an operations hire, because the standing being spent is personal and the loops being closed are personal. The instinct under fundraising pressure is to widen the ask list, which increases the draw rate on a pool that is already signalling. The cheapest first test costs one message: ask the contact who has helped you most what a normal request rate looks like at your stage. If the answer is materially below what you have been doing, you have found the reason for the slow replies without having to guess at it, and the repair is available while the relationship is still in the warning stage rather than after it.
Sources and notes
- Elinor Ostrom, Governing the Commons: The Evolution of Institutions for Collective Action, Cambridge University Press, 1990. The central empirical argument is that common-pool resources are frequently governed successfully by the users themselves, through institutions that are neither private property nor state control, and that these arrangements have in documented cases persisted for centuries. Cited here for the claim that shared-resource collapse is a failure mode rather than an inevitability.
- Michael Cox, Gwen Arnold and Sergio Villamayor Tomás, A Review of Design Principles for Community-based Natural Resource Management, Ecology and Society 15(4), 2010, article 38. Open access: https://ecologyandsociety.org/vol15/iss4/art38/main.html. The paper reviews 91 studies that apply Ostrom’s design principles and reports that the principles are well supported empirically, while recommending that several be reformulated. Cited for the specific structural features listed in section 1. Note the qualifier: well supported is not universally supported, and the authors argue for revision rather than for treating the list as settled.
- Martin W. Cripps, George J. Mailath and Larry Samuelson, Imperfect Monitoring and Impermanent Reputations, Econometrica 72(2), 2004, pages 407 to 432. The result is that in a repeated game with imperfect monitoring, a reputation for playing a strategy is temporary: beliefs about the long-run player’s type eventually converge, and the reputation effect disappears. Developed at book length in George J. Mailath and Larry Samuelson, Repeated Games and Reputations: Long-Run Relationships, Oxford University Press, 2006. Cited in section 2 for reputation behaving as a flow rather than as an accumulated balance.
- Jeffrey Carpenter and Andrea Robbett, Game Theory and Behavior, MIT Press. The common-pool resource game, its experimental record, and the conditions under which communication and sanctioning sustain cooperation are treated in the chapters on public goods and common resources. Used here for the point that observed behaviour in these games is neither full cooperation nor full depletion, and responds strongly to whether users can monitor and sanction each other.
A note on a number this article does not give. It would be convenient to state a safe ratio of asks to reciprocations. No such number transfers. The ratio that a specific network tolerates depends on how large the pool is, how fast it regenerates, and what you are worth to it, and the only reliable way to learn yours is to count your own and then ask one person whether it is high.
Joshua Agonya Pi’Rwot, Founder.