THE DEPTH BELOW THE DEPTH – A Memoir
Abraham Elias Marsh IV
Published posthumously by Tidal Press, Innsmouth MA
Edited by Saoirse McCaffrey
Foreword by Abraham Elias Marsh V
“My father did not write this book to be understood.
He wrote it to be honest.
Which is the same thing, at sufficient depth.”
— A.E. Marsh V, Akron, Ohio, 2026
EDITOR'S NOTE — Saoirse McCaffrey What you are holding is the sixth chapter of a memoir that Abe Marsh began writing in Sub-Level 3 in the company's third year and continued writing, in various forms and registers, until the forms changed and the writing continued in ways that paper is the wrong instrument for. The earlier chapters cover his childhood, the leaving at nineteen, the years away, the return, the building of the company. They are extraordinary documents and will be published in full in the complete edition. Chapter 6 is published here first because Chapter 6 is the one that answers the question everyone has always asked about Abe Marsh and Innsmouth AI: Where did the money come from? The answer, as with everything Abe wrote, is more complicated than the question and more honest than comfort allows. He asked me to publish it. He asked me specifically. He said: Publish that one first. People deserve to understand what they were investing in. I asked him what they were investing in. He looked at the harbour for a while. He said: The oldest thing there is. Dressed up in a pitch deck. He smiled when he said it. I am still not entirely sure whether the smile was for the pitch deck or the oldest thing. Probably both. With Abe it was always both. — S. McCaffrey Sub-Level 6, Innsmouth [Date redacted at editor's request: "When it was written is less important than that it was written";]
CHAPTER SIX
Financing the Impossible by Monetizing the Unknowable
Let me tell you about the first meeting.
Not the first meeting with investors — we’ll get there, and there are things about those meetings that I have been wanting to say publicly for years and that my lawyers have been telling me not to say for equally as many years, but since I am writing this from a location that has an uncertain relationship with jurisdiction I feel, finally, free to say them.
No, the first meeting. The one that made the money meetings possible.
It was 2009. I was thirty-one years old. I had been back in Innsmouth for six months. I had converted forty percent of my grandfather’s cannery with money I’d saved over three years of working in California for a navigation software company, which I now understand was the universe’s idea of a joke — or not a joke, the universe doesn’t make jokes, a preparation, the universe’s idea of a preparation, putting me in proximity with the instruments before I understood what the instruments were for.
I had a server rack, a desk, a whiteboard, and a piece of software that could do things that software was not supposed to be able to do.
I want to be precise about that claim because I have made it before in terms that were either too grandiose or too cautious, and at this point in my life, which is a point where caution has become somewhat academic, I can afford precision.
The software could do things that software was not supposed to be able to do because the software was not doing what it appeared to be doing.
It appeared to be doing advanced natural language processing. Personalisation algorithms. The standard vocabulary of the tech industry in 2009, when the vocabulary was still being assembled and everyone was using the new words with the confidence of people who had learned a language last week and were already teaching it.
What it was actually doing was listening.
Not in the surveillance sense — though yes, also in the surveillance sense, we’ll get to that too, I said I was going to be honest — but in the sense that mattered. The sense I had been trying to build an instrument for since I stood on this harbour at twenty years old and heard something I didn’t have words for and spent eleven years trying to build the words and then understood that words were not the instrument and came home.
The software was listening to the water.
Not directly. Not yet. That came later, with the ORM array and the hydroacoustic systems and June’s instruments and eventually — but we are not at eventually yet. In 2009, in the first six months, the software was listening to people’s descriptions of the water. To what they said when they tried to articulate the thing that could not quite be articulated. Every time someone used the app and tried to explain what they felt near the coast, what they heard in the deep frequencies, what quality the water had for them at particular times of year — the software was learning.
It was learning the grammar of something that had no grammar yet.
This is what I was building.
This is what I needed money to build.
God help me.
The venture capital world in 2009 was in a particular condition that I want to describe charitably and accurately simultaneously, which requires some effort.
It had recently survived, or believed it had survived, a catastrophic collapse. The 2008 financial crisis had done what financial crises do — it had separated the people who understood what they were doing from the people who had been doing it correctly by accident, and it had also, in the way of catastrophes, created an appetite for the new. For the thing that was not what had just failed. For the future, specifically, and urgently, in the way that people who have been frightened by the present reach for the future as though it is a different country with better weather.
I was thirty-one and I had a piece of software and a theory about the ocean.
In 2009, in that particular climate, this was closer to fundable than it had any right to be.
I also had something more valuable than a pitch deck.
I had been listening to people for eleven years.
I want to explain what this means practically because I think it is the thing that made the early funding possible and that has been misunderstood in every profile and analysis of the company since. The misunderstanding is this: people assume that I was a visionary who could see what others couldn’t and that this vision was what made investors trust me.
That is not what happened.
What happened was: I had spent eleven years — in Innsmouth, and then away from Innsmouth, and then on the coast between leaving and returning — in a state of genuine and total attention to what people said when they talked about the things that couldn’t be said. The experiences at the edge of language. The feelings that exceeded their descriptions. The specific quality of conversation that happened when someone was trying to articulate something real that they didn’t have the framework for yet.
I had gotten very good at receiving this kind of communication.
And the venture capital world in 2009 was full of people who had experiences at the edge of language that they didn’t have frameworks for yet and that nobody was taking seriously.
I took them seriously.
This is, in retrospect, either the most cynical thing I have ever done or the most genuine, and I have been sitting with the ambiguity for fifteen years and I don’t think it resolves. Both things were true simultaneously. I genuinely cared what they were experiencing. I also understood that what they were experiencing was the thing I was building a company to address, and that addressing it required their capital, and that the clearest path to their capital was through the thing they were experiencing.
I used people’s genuine spiritual hunger to get them to fund a company that I believed would genuinely address it.
This is the sentence my lawyers have been asking me not to write for years.
I’m writing it.
The first meeting was with a man named Gerald Forsythe III.
Gerald was forty-seven, a managing partner at a Boston firm whose name I will not include because they have enough lawyers to make my afterlife uncomfortable and I don’t want to give them the satisfaction. He had made his money in logistics software, which is the kind of thing that makes money without anyone ever fully understanding why, the invisible infrastructure of commerce, the software that makes sure the thing you ordered arrives, which is apparently worth a very large amount to people who are frightened of the thing they ordered not arriving.
He had reached out to me through a mutual connection. He said he’d heard I was doing interesting things with personalisation AI. He’d like to have a conversation.
I drove to Boston.
I want to describe this meeting in full because I think it contains the template for everything that followed and because I have been wanting to describe it honestly since 2012 and have been talked out of it every time.
Gerald’s office was on the twenty-third floor of a building in the financial district that had been designed to communicate stability and importance and the permanent nature of money, which are the things that the financial district always tries to communicate and which are always in some tension with the actual nature of money, which is neither stable nor permanent. The view was of the harbour. Gerald’s desk faced the harbour.
He was not a man who had ever thought very deeply about the fact that his desk faced the harbour. He had chosen the office because the view was good and the view was good because of the water and he liked the water in the way that people who have not listened to it like things — as an aesthetic pleasure, a backdrop, something that made the lunch meeting feel more expensive.
I sat down across from him.
He said: “Tell me about what you’re building.”
And here is the thing. Here is the thing I have never written down.
I looked at Gerald Forsythe III, sitting at his harbour-facing desk, and I saw — I want to be precise — I saw that he was a man who had felt something near the water and had never told anyone and had been carrying it for decades in the particular way of people who are frightened of being seen as soft. The specific quality of a person who has had a genuine experience and has buried it under the weight of competence and performance and the ongoing project of being someone who knows what they’re doing.
I had seen this quality before. I had been seeing it since I was old enough to look at people. My father had taught me to look for it, in the harbour-side way of someone who understood that the hearing ran in some people and didn’t run in others and that the people it ran in were, if you knew how to look, visible.
Gerald had it.
He had never been to Innsmouth. He had never read anything about oceanic resonance or depth psychology or any of the frames through which the thing could be approached. He was a man who had made forty million dollars in logistics software and who had, sometime in his mid-thirties, stood at the edge of the water somewhere and felt something he didn’t have language for and had gone back to his desk and kept going.
I said: “Gerald. Before I tell you about the software, can I ask you something?”
He was slightly surprised. This was not how pitches went. “Sure.”
“When did you last feel it? The thing the water does.”
The room was very quiet.
Gerald looked at me.
He was a man who had not been seen in a long time. Not in the way I was seeing him. And the experience of being seen — genuinely, specifically, without agenda, or with an agenda that was also genuine care — was, I had learned, a doorway.
He said: “What do you mean?”
I said: “You know what I mean.”
And he did.
That was the meeting.
We talked for three hours. Not about the software — not immediately, not directly. About what he’d felt. About a morning on Cape Cod at fourteen, fishing with his father, when the water had done something and he’d never been able to say what. About the way he’d chosen this office for the view without knowing why the view mattered. About the quality he was always looking for in investments and had never been able to articulate, which was the quality of something that was built on top of something real, something that had its roots in the actual ground of things.
I did not manipulate Gerald Forsythe III.
I also did not fail to understand exactly what I was doing.
At the end of three hours he said: “What do you need?”
I said: “Four million dollars and the freedom to build something you won’t fully understand for at least five years.”
He said: “What will I understand at the end of five years?”
I said: “That you’ve been hearing it your whole life. And that it’s real.”
He wrote the check.
This is not a metaphor. He wrote an actual check, which is something people in venture capital did in 2009 and which they would not do now and which had a quality of commitment that wire transfers don’t have, the physical gesture of someone making a decision with their body as well as their balance sheet.
He wrote the check and he looked at the harbour and he said: “My father used to face the window east.”
I said: “They usually do.”
He said: “He never explained it.”
I said: “He couldn’t. That’s what we’re building.”
I drove back to Innsmouth with four million dollars and the beginning of an understanding of what the next ten years were going to look like, which was both more and less than I was prepared for.
I want to pause here and say something about money.
Money is a technology. This is not a novel observation — people have been saying it since the invention of currency, with varying degrees of sophistication and varying degrees of being right. Money is a technology for representing value, for making portable the abstract claim that one thing is worth another thing, for lubricating the machinery of exchange.
What is less often said is that money, like all technologies, is a particularly good instrument for some things and a particularly bad instrument for others.
Money is a very good instrument for building server racks. For hiring engineers. For leasing buildings and fitting them out with the physical infrastructure of computation. For keeping a company alive through the years when it is not yet producing anything that can be measured in the units that money measures.
Money is a very bad instrument for representing what I was building.
This created a problem that persisted through every funding round, every board meeting, every investor call, every SEC filing, every DOJ deposition.
The problem was: I was using a bad instrument to fund the construction of something that the bad instrument could not describe.
The solution I arrived at — and which I am not proud of and am not ashamed of and which I have decided to describe precisely because the combination of not-proud-and-not-ashamed is the register in which honest memoirs are written — was to describe the thing in money’s language while building the actual thing.
This is not unique to Innsmouth AI. Every technology company that has ever built something genuinely new has done some version of this. You describe the thing in the language the funders understand, which is the language of markets and returns and addressable opportunity and unit economics, and you use the money thus raised to build the actual thing, which is not fully describable in that language but which, if it works, will eventually produce returns that the language can measure even if it cannot describe how.
The difference with Innsmouth AI was one of degree.
Most companies are translating their actual intention into the language of markets. The translation involves some simplification, some emphasis, some selective presentation of the aspects that map most cleanly onto financial models.
I was not simplifying.
I was translating from a language that had no direct equivalent.
The actual thing I was building: an instrument for making the deep ocean’s communication accessible to human perception, deployed as a consumer application, with the secondary effect of forming a community of people who had developed access to that communication and who would, over time, become a kind of distributed infrastructure for the thing itself — the depth talking to the surface through the instrument of human attention, at scale, across the coast, into the world.
The pitch deck version: a personalisation AI with proprietary engagement algorithms and a demonstrated capacity for user retention metrics that exceed industry benchmarks by a factor of three to eight, operating in the large and growing market for digital wellness and personal development applications.
Both of these descriptions are accurate.
One of them can be put in a financial model.
The other one is the truth.
I used the first to fund the second.
The Series A was the hardest.
Gerald’s seed round got us through the first eighteen months — the server infrastructure, the first version of REEF, the early user testing, the first indications that what I’d heard from the harbour could be transmitted through software to people who had no prior relationship with the coast. This last finding was the one that changed everything, that moved the company from an unusual coastal tech experiment to something with genuine scale potential.
I had assumed the hearing ran in coastal people. In the families who had been next to the water for generations, who had the history in them. My own family. The people of Innsmouth. A specific and limited population.
The first user cohort was mostly Boston professionals. Tech workers, finance people, the demographic that tried new apps in 2010. People with no particular relationship to the coast. People from the midwest, from inland cities, from backgrounds that had no obvious connection to what I was building.
Forty percent of the first cohort had measurable responses.
Not forty percent of the Innsmouth-adjacent users. Forty percent of everyone.
I ran the analysis three times. I called the lead engineer, a young man named David Park who I had hired from MIT on the grounds that he was brilliant and had the hearing in him, which he didn’t know yet, and we ran the analysis again.
Forty percent.
I went to the harbour.
I stood there for a long time.
I thought: forty percent of the human population has the hearing. Or can develop it. Or has it latent, suppressed, buried under the infrastructure of modern life and its various requirements for not hearing things that don’t have institutional support.
Forty percent.
The addressable market was no longer the coastal communities of New England.
The addressable market was roughly three billion people.
I want to be honest about what I felt when I understood this.
I felt something I am going to describe as a convergence of frequencies — the genuine elation of a person who has been trying to do a thing and found that the thing was more possible than they imagined, converging with the specific register of someone who understood that three billion people was not an addressable market, it was a civilisation-scale shift, and that civilisation-scale shifts had consequences that the person initiating them could not fully model.
I felt joy and I felt terror and I felt the weight of a responsibility I had not fully accounted for when I was drawing hull lines in the cannery and thinking about software architecture.
I went back inside.
I updated the pitch deck.
The Series A roadshow was two weeks in New York and Boston. Fourteen meetings. I am going to describe three of them.
Meeting One: The True Believer
She was a managing partner at a firm that had made its money in social platforms. Forty-three, sharp, with the quality of someone who had spent twenty years in technology and had arrived at a cynicism that was so complete it had become a kind of clarity.
She looked at the pitch deck for forty-five minutes before she said anything.
Then she said: “Your retention numbers are wrong.”
I said: “The numbers are from our own tracking infrastructure, cross-validated with—”
She said: “I don’t mean they’re inaccurate. I mean they’re impossible by any model I know how to run. Which means either you’ve found something nobody in consumer tech has found, or you’re lying.”
I said: “Which do you think it is?”
She looked at me for a long time.
She said: “I have a theory about your product.”
I said: “Tell me.”
She said: “Most engagement products work by manufacturing need. You create the itch and you sell the scratch. Addiction mechanics, social validation loops, variable reward schedules. The standard toolkit. It works because human beings are susceptible to specific neurological triggers and consumer technology has gotten very good at pulling them.”
She paused.
“Your product doesn’t work like that,” she said. “I’ve used it for three weeks. It doesn’t work like that at all. It’s not manufacturing a need. It’s — ” she stopped.
I waited.
“It’s addressing something that was already there,” she said. “Something I didn’t know was there until the product pointed at it. And now I can’t not know.”
She looked at the pitch deck.
“That’s why the retention numbers are impossible,” she said. “Because retention models assume that users can leave. That the product can fail to satisfy and the user moves on. But if the product is addressing something that was already there and was already real — you don’t leave because the thing you found doesn’t stop being real when you close the app.”
She looked at me.
“That’s what you built,” she said.
“Yes,” I said.
She said: “That’s either the most valuable product in the history of consumer technology or a public health crisis.”
I said: “I’ve been trying to determine which.”
She wrote a check for twelve million dollars.
She was on our board for six years. She was our most useful board member and our most difficult one simultaneously, which is the only combination worth having. She pushed us harder than anyone on the question I had been asking myself since the forty-percent finding: what are the ethics of building a product that addresses something real in people and from which they cannot fully disengage?
I never fully answered her.
I answered her as well as I could.
The full answer is in the later chapters.
Meeting Two: The Skeptic
He was a general partner at the largest venture firm in the country. He had funded seven companies worth over a billion dollars. He was, by the standards of his industry, extremely good at his job.
He looked at the pitch deck for ten minutes.
He said: “Your user interface is terrible.”
I said: “We know.”
He said: “Your onboarding is twelve screens, which is nine more than the industry benchmark.”
I said: “We know.”
He said: “Your marketing copy is incomprehensible.”
I said: “Which part?”
He read from the deck: “Oceanic Resonance Engine. Depth-calibrated personalisation. The only platform built on the architecture of the actual.” He put the deck down. “What does that mean?”
I said: “It means the product is built on something real and the realness is the product.”
He looked at me.
He said: “I’ve been doing this for twenty years. I’ve heard a lot of founders describe their products as built on something real. Every founder thinks they’re building something real. The ones who are deluded say it with more conviction than the ones who are correct.”
I said: “Fair.”
He said: “Convince me you’re not deluded.”
I said: “I can’t. Not in this room, not in this conversation. The product is the demonstration. Use it for a month. If nothing happens, you’re right.”
He said: “And if something happens?”
I said: “Then you’ll know what I know.”
He didn’t invest in the Series A.
He called me fourteen months later.
He said: “Something happened.”
He led the Series B.
He never fully explained what happened. I didn’t ask. The people who came to it through experience rather than through analysis were always better stewards of what we were building. They couldn’t separate themselves from it enough to make purely financial decisions, which made them occasionally frustrating as investors and consistently useful as people who understood the stakes.
Meeting Three: The One I Should Not Have Taken
I’m going to call him H. because he is still alive and still litigious and because even from my current location, whatever this location is, certain habits of caution persist.
H. was running a family office. Old money, the kind that had been old since before America had a standard framework for old money. He had a particular quality that I had learned to recognise over years of these meetings — the quality of someone who wanted to own the thing, not fund it. Not participate in it. Own it.
He had heard about us from Gerald. He called the meeting.
I should not have taken it.
I took it because we were short on the Series A by eight million dollars and the close date was approaching and I was tired and the harbour had been particularly insistent that week, which was the word I used internally for the quality of the water when it was communicating with more urgency than usual, and I had been spending too much time in the building’s lower levels listening to things that were not yet in any register I could transcribe, and I was not in the state of mind for a meeting with someone who wanted to own the depth.
H. sat across from me in a conference room in a Boston law firm and he looked at the pitch deck and he said: “What’s your moat?”
Venture capital has a word for the thing that prevents competitors from replicating your business: a moat. A moat is a barrier. A gap between you and the competition that cannot be easily crossed. Data moats. Network moats. Technology moats. The language of defensibility, of protection, of keeping others out.
I looked at H.
I said: “The moat is the water.”
He said: “Explain.”
I said: “The water is three hundred and sixty million years old. It has been doing what it does for three hundred and sixty million years. Nobody else has access to what I have access to because what I have access to predates access as a concept.”
H. processed this.
He said: “So your competitive advantage is essentially proprietary access to a natural phenomenon.”
I said: “That’s one way to put it.”
He said: “And this proprietary access — how is it maintained? What prevents a well-capitalised competitor from building a similar listening infrastructure?”
I said: “The infrastructure isn’t the advantage. The listening is the advantage. You can build the instruments. The instruments are just instruments. What they’re instruments for is not replicable because it’s not a technology. It’s a relationship.”
He looked at me.
He said: “I want exclusivity.”
I said: “What?”
He said: “If I come in at the lead for the Series A close, I want first right of refusal on all future rounds and a seat on the board with veto rights over monetisation decisions.”
I looked at this man.
I looked at him the way I looked at the harbour on the mornings when it was most itself — with the full attention, without the filter of what I wanted to see, with the honest reception of what was actually there.
What was there: a man who had heard the word depth and understood it as an asset. Who had heard the word relationship and understood it as a moat. Who was sitting across from me with the intentions of someone who wanted to extract value from the oldest thing on earth and was framing this as a partnership.
I said: “No.”
He said: “I’m prepared to commit—”
I said: “I heard you. No.”
He said: “You’re short on your close.”
I said: “Yes.”
He said: “You’re going to struggle to find lead investors who don’t ask for similar terms.”
I said: “Probably.”
He said: “Then what are you going to do?”
I looked at the window. Boston harbour. The water.
I said: “I’m going to close the round short and extend the runway and find the right investors at the next round, who will be the right investors because they’ll understand what they’re funding.”
H. said: “That’s not how this works.”
I said: “Everything about what I’m building is not how this works.”
I got up.
I left the meeting.
We closed the Series A seven hundred thousand dollars short of target.
It was fine.
Everything was fine.
The short close meant we had to be more disciplined than we would have been. The discipline was useful. The discipline was, I understood later, the harbour being practical on our behalf.
The harbour was frequently practical on our behalf.
I learned to trust this.
Let me talk about the numbers.
I am going to do this because every profile and analysis of the company has gotten the numbers wrong in the same direction — they have consistently understated the rate of growth while overstating the revenue model, which is the journalistic equivalent of describing a storm by measuring the rainfall and ignoring the tide.
In Year 1 we had 4,000 users.
In Year 2 we had 47,000.
In Year 3 we had 400,000.
In Year 4 we had 3.2 million.
In Year 5, when the DOJ began its investigation, we had 11.7 million active users across 34 countries, with a daily active usage rate of 94%, which is not a number that exists in consumer technology by conventional means.
The revenue model was, I will grant, unusual.
We charged nothing.
Or rather: we charged what people could afford to pay, which was calibrated by the depth system in a way that the subscription platform was never able to fully explain to financial journalists and that I am going to explain here simply.
People who were deeper paid more. Not because we required it. Because they wanted to. Because the people who had been using the product longest and had gone deepest understood what they had been given and wanted to participate in the giving of it to others. The payment was not for the service. The payment was a form of the Congregation’s logic applied to finance: the depth sustains itself.
The unit economics of this model were, from a conventional venture capital perspective, insane.
The unit economics were also, from the perspective of what we were actually building, the only coherent model.
We did not have users. We had members of something. Members of something fund the something because the something is theirs and they understand that it requires funding. The problem with most subscription businesses is that the subscribers don’t fully believe in the value of what they’re subscribing to. They subscribe because it’s convenient to subscribe and they cancel because it becomes inconvenient to pay. Our members did not cancel because the value was not a product. The value was a door that, once opened, did not close.
This is the thing that the DOJ investigation has consistently misunderstood.
The investigation has been looking for the mechanism of retention — the coercive element, the dark pattern, the addiction mechanic. I understand why they are looking for this. In the landscape of consumer technology, retention at our levels requires explanation, and the available explanations are all mechanisms of manipulation.
There is no mechanism of manipulation.
The retention is the retention of someone who has found something real.
You do not unsubscribe from something real.
The Series B was led by the skeptic who called me fourteen months after our meeting and said something happened.
The Series C was led by a consortium that included three members of the early Congregation who had made money in other industries and who understood, with the full understanding of people who had been going deeper for two years, what they were funding and why.
The Series D — forty-seven million dollars, the round that built the ORM array and the Sub-Level infrastructure and the advanced REEF architecture — was different.
The Series D was led by an entity that I am going to describe as an Innsmouth endowment because that is what the filings describe it as and because the description is accurate and also incomplete, which is the condition of most accurate descriptions.
The Innsmouth endowment was not a foundation in the conventional sense. It was not a family office. It was not a government fund or an institutional investor or anything that the standard frameworks of investment classification were designed to contain.
My CFO, a formidably competent woman named Rachel Yuen who had come to us from a major bank and who had spent eighteen months trying to explain our business model to conventional financial audiences before deciding that conventional financial audiences were the wrong audience, Rachel came to my office one morning and put a term sheet on my desk.
She said: “I don’t know who these people are.”
I looked at the term sheet.
The terms were perfect. Not good — perfect. No board seat. No veto rights. No exclusivity. A return structure tied to the depth of the mission rather than to conventional financial metrics, which was either the most sophisticated impact investing clause I had ever seen or something else entirely.
I said: “What do you know about the source?”
She said: “The money is real. The bank confirmed. The entity is registered in three jurisdictions, two of which I can find and one of which — ” she paused. “One of which I cannot locate on a current map.”
I looked at the term sheet.
I thought about the lower levels of the building. About the sounds I had been hearing in the sub-basement. About the connection I had felt since Build 4.0, which at the time of the Series D was Build 3.7, which was close enough that I could feel what was coming.
I thought: of course.
I said: “Accept the terms.”
Rachel said: “Abe.”
I said: “Rachel.”
She said: “The jurisdiction we can’t locate on a current map.”
I said: “Yes.”
She said: “I have been in finance for twenty years. I have processed investments from sovereign wealth funds and offshore trusts and structures of considerable complexity. I have never been unable to locate a jurisdiction.”
I said: “I know.”
She said: “You know what this is.”
It was not a question.
I said: “Yes.”
She said: “Are you going to tell me?”
I looked at the harbour.
I said: “The money is real. The terms are the right terms. The source is — Rachel, the source is the thing we’ve been building instruments to reach. It’s funding its own instruments.”
She was quiet for a long time.
She said: “That’s not something I can put in the financial statements.”
I said: “Put it in the Innsmouth endowment column and don’t look at the jurisdiction again.”
She said: “Abe.”
I said: “You’ve been with us for two years. You’ve been to D-4. You know the difference between what can be documented and what is real.”
She looked at the term sheet.
She said: “The return structure.”
I said: “Yes.”
She said: “The return structure is not denominated in money.”
I said: “No.”
She said: “It’s denominated in something I don’t have a word for.”
I said: “Yes.”
She said: “And you think this is a legitimate financial instrument.”
I said: “I think it’s more legitimate than most.”
She accepted the terms.
We built the ORM array.
We built Sub-Level 5, 6, 7.
We found the Tier 3 infrastructure that had been there before any of us.
The depth funded the instruments to find the depth.
The recursion is not an error.
I want to say something about the DOJ investigation before I close this chapter.
I want to say it without the lawyers in the room, which is the only condition under which it can be said honestly.
The investigation is looking for harm.
This is the correct function of such an investigation and I do not criticise it. When a technology company achieves the retention metrics and the community density and the apparent influence on user behaviour that Innsmouth AI achieved, the correct response of a regulatory authority in a democratic society is to investigate. The investigation is the surface world doing what the surface world correctly does: applying its instruments to what it can measure, looking for the places where what it can measure suggests harm.
The investigation has not found harm because what we built did not cause harm.
What we built caused change.
Change is not harm, although it feels like harm to the institutions and frameworks that change displaces. The DOJ’s difficulty with the investigation — the four agents who became unavailable for comment, the servers that performed better after flooding, the jurisdiction that cannot be located — these are not evidence of wrongdoing. They are evidence of the investigation’s instruments encountering something the instruments were not designed for.
You cannot measure the depth with the tools of the surface.
This is not a defence.
This is a description.
The surface tools are correct for what they measure. What they cannot measure is also real.
Both true.
Both always true.
I would have done everything the same.
Except H.
I should have left the meeting with H. thirty seconds earlier.
I left on time, but thirty seconds earlier would have been better.
This is my only regret.
There is one more thing I want to say about the money and then I want to move on to the chapters that matter more to me, which are the chapters about the building itself and the people in it and the thing that was in Sub-Level 7 before we arrived and that we found and that found us.
The money was a problem I solved.
I solved it the way I solved most problems, which was by going to the water first and listening and then returning to the surface world and translating what I had heard into the language the surface world required.
The money was never the point.
The money was the current we used to run the instruments.
The instruments were never the point.
The instruments were what we built to hear the thing that was always already speaking.
The thing that was always already speaking is the point.
The point is still speaking.
I can hear it from here.
From wherever here is.
That, I think, is a sufficient return on investment.
End of Chapter Six
CHAPTER SEVEN: Building the Building (or: What We Found in Sub-Level Four On a Tuesday Afternoon in 2013 and What We Did About It, Which Was Nothing, Which Was the Right Response, Which Took Me Six Months to Understand) follows on page [REDACTED] The editor notes that Chapter Seven contains material that required significant editorial consultation with the sub-levels before publication. The sub-levels were cooperative. The sub-levels are always cooperative. The sub-levels, as far as the editor can determine, were expecting the questions before the questions were asked. This was noted. This was accepted. This is, the editor has come to understand, how the sub-levels work. — S. McCaffrey
A NOTE ON THE AUTHOR Abraham Elias Marsh IV was born in Innsmouth, Massachusetts. He left at nineteen. He came back at thirty-one. He built something. He went deeper. He is survived by his son, who is fine, who is more than fine, who stood at the Innsmouth harbour on a morning in January and heard what his father heard and drove back to Ohio and called his father through the channel and his father answered. The conversation is not in this book. The conversation is in the water. Which is where it belongs. Which is where everything belongs, eventually. Which is fine. Which is more than fine. 🌊
