A Theory of Lossless Innovation

Does Modern Work just Evaporate?

"We need better theories." --
Clayton Christensen
As the Internet, Computing, and Mobile came into maturity, the message from capital markets became clear: A select few technology startups seem to grow at exceptional speed with very little capital expenditure beyond hiring rooms full of smart twenty somethings. Of course,  they must be pampered with ping pong tables and free lunch. Onlookers often have trouble making sense of why and how these organizations grow at this kind of speed with so few employees.
I have played some role as a go-between from the startup world to the rest of the world, as I spent about 12 years in Venture Capital, somewhere before and after as a startup cofounder. A very common experience in Venture Capital is being asked to pontificate, present, or participate on a panel, with the topics all dancing around the outside world asking for some sense as to why, say, AirBnB is more valuable than most hotel brands. Why is some four year old company raising capital at valuations higher than most well-respected brands in the market? Multiply that challenge to help sense-make many times per week over 12 years, we come up with ways of thinking and talking about things.
Spoiler: startup success can be boiled down to how they collaborate, including the tools, techniques, and principles used in collaborating. One of these principles is to pursue platform-based business models, where their collaboration is not just internal or hierarchical, they are collaborating with partners, customers, and users across a distributed network of shared incentives and reliable behaviors. So, hyper-growth startups have found breakthroughs in models, tools, and techniques for collaboration.
The more exciting insight is, ironically, just how bad everyone everywhere seems to be at long-term collaboration in business, and just how imperfect even the highest growth startups are. They are often, at best, marginally better at a few ways of collaborating than the more mature industry participants, and they are often comically worse than incumbents at most other ways of collaborating. In other words, they are doing a few things brilliantly, but for most other things businesses need to do they are a fiasco.
(You may have exposure to parts of the world where this may not be the case: Skyscrapers do get built, surgeons save lives in surgery. Clearly, collaboration works in pockets. I speak of technology startups.)
This got me thinking and reflecting and discussing collaboration and productivity and value creation across different business at different stages, and suddenly I had a sixth-sense moment of "I see dead people!" -- most people I know are brilliant but commonly have the experience of working on something for a long time only to have it not go anywhere. We make presentations that get us positive feedback, then nothing happens. Projects get stalled or blocked for months on end. We spend hours writing an email that goes out and people barely skim it. We talk to 100 potential customers to get one, and then don't find ways to properly engage the 99 we started a conversation with. We launch an idea that doesn't get traction. We make documents that no one looks at ever again, and when we need it later we have trouble finding it. Most modern humans report spending hours a day just triaging emails. Even among the most highly productive people at the most highly productive companies, we are mainly doing things that turn out to not really be a good use of time. It seems like everywhere, when viewed from the long-run and not the day-to-day grind, our productivity just evaporates.

AI will just create more of everything

AI has exploded into the world at gobsmacking speed. With OpenAI's release of GPT3 in 2022, by 2025, 1/4th of all funded startups claim to be an AI startup. There is something about the potential for AI and perhaps the memory of how profoundly the rise of Search Engines, Marketplaces, and Ad Technology (including Facebook, et al.) dramatically transformed business, that it seems every business in the world is intrigued to try to use AI.
So, for older, larger organizations, organizations that at this point may have legacy systems and processes, knowing many amazing tools exist is not enough. Promoting the individuals who moonlight to figure out how to amplify their own productivity with AI does not, on its own, meet the broader challenges of management and leadership.
Organizations need an overarching strategy that can guide smaller decisions — to adopt or avoid, guidelines on how to use new tools, and animating goals that align everyone involved.
And above all that, organizations need a unifying motive and story. Tactical improvements are more valuable as part of a larger strategy, and any strategy is more valuable as part of a larger set of beliefs and animating narrative. To have beliefs that will endure beyond the day to day dynamics, organizations need to embrace a common theory.
We propose A Theory of Lossless Innovation. What do we mean? Well, let us ground it in a story, one that is a parable for the next century.

A Parable of Directionality

On June 2, 1875, Alexander Graham Bell heard the voice of his colleague, Thomas Watson, from an unfinished contraption on his desk. Watson had been tinkering with his end of the same contraption, built as yet another one of their attempts at creating an "acoustic telegraph". Bell was barely able to make out a human voice. The sound was filled with static, barely audible, but Bell instantly recognized this was the breakthrough they were seeking. This moment sparked a full century of innovation, research, engineering, capital investments, wealth creation, and public policy. Bell and Watson would have no way to understand the spark they had lit. Nine months of tinkering later, on March 10, 1876, Bell placed the first intentional call to Watson, in another room. "Mr. Watson, come here – I want to see you."
ℹ️Information
What goes in one end should come out the other.
From a certain vantage point and with a wide but focused lens, the arc of the history from 1875 on can be fit into a story about the relentless, collective, global pursuit of Lossless Communication.
Developing Bell's invention into something that could deliver audible, clear calls, reliably and over any distance would take a century. The endeavor involved an astounding number of people, and countless inventions and technologies that had direct application to the goal at hand. From Information Theory to Cryptography, from copper wires to fiber optic cables, from switchboards to transistors, one animating story united a vast network of institutions, companies, scientists, engineers, banks, stock markets, and shareholders. No matter what their day to day, or their year to year, or this organization or that, they were all engaged in one pursuit: Lossless Communication.
Even more worth understanding is that the countless inventions and technologies that had no direct, immediate commercial application would go on to reinvent the whole world. Bell Labs, which housed the research and development arm of the near monopoly that was AT&T at its zenith, would also spin out
  • the foundations of computer-based cryptography.
  • the photovoltaic cell, the pixel of solar energy.
  • charge-coupled devices, the base of digital photography and radio astronomy
  • the C programming language, still in use today for any software that needs fast-execution systems programs.
  • operating systems, the software that manages computers. UNIX, developed at Bell Labs, heavily influenced MacOS and Linux.
  • cellular networks for mobile device calls and data
  • lasers, which are now the primary instrument in lithography to make chips, fiber optic data transmission, realtime computer vision, untold military applications in precision weapons, and various medical procedures.
  • foundations of neural networks and machine learning, such as Support Vector Machines
Bell Labs was the epicenter of invention in the 20th Century. An eager read of its hagiography, The Idea Factory, or even just a perusal of its Wikipedia page, inspires awe. Work at Bell Labs led to eleven Nobel Prizes and five Turing Awards. Their inventions defined most of the modern era. The Transistor, the Laser, photovoltaic cells, charge-coupled devices. The Unix operating system, the C, C++, and AWK programming languages. The field of Information Theory was birthed at MIT, but Claude Shannon, the creator, worked at Bell Labs for much of his life.
From the animating, common goal of Lossless Communication, we have most technologies that define life in the 21st Century.

Lossless as Analogy

ℹ️Information
What goes in one end should come out the other.
The term Lossless has become technical, referring to algorithmic methods that reduce then reconstruct data with perfect fidelity. I posit that applying it as an analogy to almost any human pursuit will loosely predict the emergence and distribution of invention, technology, and commerce.
Here is the pattern I want to name. When a technology is first invented, it is lossy. The first telephones carried a voice only barely, and only from one room to the next. The first photographs took hours to expose and could not be copied. The first computer networks moved a trickle of data and fell over often. In every case the invention arrives clunky and inefficient, with most of its market value still trapped inside it, unrealized.
What follows is not one breakthrough but decades of them. For thirty years or more, whole industries organize themselves around a single quiet imperative: make sure that what goes in one end comes out the other. Close the gap between what was sent and what was received, between what was captured and what was reproduced. Each increment of recovered fidelity opens a new market, funds a new firm, and employs a new profession. The value does not arrive with the invention; it is squeezed out of the invention, a little at a time, over a generation.
The arc really runs in three stages: Noise → Lossy → Lossless. At first there is only noise — the signal exists in principle but cannot be recovered at all; the very first telephone call was, quite literally, mostly static. Then comes the lossy stage, where some of the signal gets through and most of it is thrown away. Then the long climb toward lossless, where what goes in one end reliably comes out the other. This progression — from noise, through lossy, toward lossless — is less a law than a narrative, in the way Moore's Law is a narrative. Moore's Law was never physics. It was an expectation, a story precise enough that an entire industry could align its investment, hiring, and roadmaps to it for half a century, and in aligning to the story, made it come true. "Lossless" can work the same way. It is the direction a network of otherwise-unrelated people — scientists, financiers, engineers, regulators — can point at and agree on without ever meeting. Directionality, not a destination.
The arc has a second axis, and the two run together. As a technology gets more lossless, it also gets more accessible — cheaper, simpler, and available to more people. Photography began with a single plate that took Maxwell hours to expose in 1861; it passed through Kodak's promise, "you press the button, we do the rest"; it now sits in a camera in every pocket, taking images of a fidelity Maxwell could not have imagined, for free, everywhere. The telephone made the same journey, from Bell's bench to a handset in every home to a supercomputer in every hand. Fidelity and access climb the same ladder: the squeezing that makes a technology lossless is the same squeezing that makes it cheap enough for everyone.
And here is the twist worth sitting with. The value does not stay in the domain where the squeezing happens. In the century-long pursuit of Lossless Communication, the byproducts — the transistor, the laser, the C language, UNIX, the photovoltaic cell — reinvented far more of the world than the telephone itself ever did. Chasing fidelity in one narrow domain turns out to be among the most reliable engines of unrelated invention we have. The case studies below are all the same arc, run in different materials.

Other Historical Case Studies

Imagery

The first color photograph is attributed to James Clerk Maxwell, the image taken in 1861.
Kodak and Xerox were also pursuing Lossless goals, pursuing the goal of making images better and better, and more and more widely accessible.
At age 24, George Eastman was planning a trip to the Caribbean for the summer of of 1878. He wanted to keep his memories. Hauling around a wet plate camera, taking pictures, and developing photos was still an unwieldy process. Two years later, Eastman received a patent from the US Patent Office, patent 226,503, for a "Method and Apparatus for Coating Plates for Use in Photography." [1]
After selling the first roll film and hand camera in 1888, George Eastman founded Eastman Kodak. Kodak went on to become the market leader in photography. It's no wonder, then, that Xerox — founded in the same city, Rochester — became the market leader in photocopying.
Steve Sasson was an engineer at Kodak when he invented the world's first digital camera in 1975. Willis Adcock was working with Texas Instruments when he filed a patent for a digital camera in 1977. [2]
Xerox developed PARC in 1970, based near Stanford University in Palo Alto, CA as it pursued Lossless image duplication. PARC follows Bell Labs closely in the sheer number of inventions that have managed to reinvent our world. Wikipedia lists major inventions as:
  • the Personal Computer,
  • the Graphical User Interface,
  • the Ethernet,
  • Object-Oriented Programming,
  • the Mouse,
  • VLSI design for semiconductors.

Sound

On the success of AT&T, many brilliant inventors were searching for a wireless method of transmitting data. Guglielmo Marconi was the first to crack it, sending a transmission 1.5 miles, or 2.4km, in 1895. By 1897, he was sending signals 34 miles, 55km, within England. By 1899, he sent signals across the English Channel. In 1900, he filed his famous patent and renamed his corporate entity Marconi's Wireless Telegraph Company. In 1909, Marconi was awarded the Nobel Prize in Physics.

Data

DARPA did not have a vision for the Internet. As America entered the Nuclear Age, the American Defense establishment became paranoid that a single bomb could wipe out all of their data. DARPA published its usual challenge to the public: invent a way to network computers so we could make data redundant.
Multi-Modal Databases

Memories

Facebook and LinkedIn could be thought of as networks of humans trying to keep (not lose) their relationships or their memories. For a long time, Facebook was just a way to share photos and "tag" the people featured in them.

Human Nature

Similar to Mimetic Theory, to pursue Lossless goals is part of human nature itself. Decades of research in Behavioral Economics have demonstrated that Loss Aversion is one of the most powerful of human motives.

The Second Track: Lossless Knowledge Work

There is a second way to run this analogy, and it is the one that matters most for the century ahead. Turn the lens away from telephones and cameras and point it at the work itself — at knowledge work, the thing most of us now do all day.
We have been collaborating in offices for over a century. And yet: some companies are productive and some are not, some can innovate and some cannot, some can execute and some only flail. Why? The honest answer is that our knowledge work is lossy. Vast fractions of it never come out the other end.
Here is the strange part. Every office looks the same — rows of desks, backs of heads, faces lit by screens. From the outside you cannot tell a productive company from a wasteful one. But underneath that identical surface, each organization has made wildly different, almost entirely invisible choices about how to organize the work so that value is captured rather than lost. The difference between the good and the bad is enormous, and you would never know it by looking.
%% TODO: insert the 70–80% knowledge-work-loss diagram here %%
So what if most of it is simply lost? Not ten or twenty percent — what if seventy or eighty percent of knowledge work evaporates? The presentation that earns a nod and then nothing. The document no one opens again and no one can find when they finally need it. The ninety-nine conversations you started in order to win one customer. The hours a day spent triaging email. Viewed from the day-to-day it feels like progress; viewed from the long run it is loss. Our offices are lossy the way the first telephone was lossy — most of the signal thrown away.
This reframes what AI is actually for. The loud story is that AI is "Generative" — a machine for producing more: more text, more images, more code, more of everything. But if knowledge work is already lossy, then generating more into a leaky system just spills more on the floor. The deeper opportunity is not to generate more; it is to lose less — to use AI to fundamentally reorganize knowledge work the way the factory once reorganized the workshop.
Because that is what today's office still is: a workshop. Every knowledge worker is an artisan, doing a little of everything in their own idiosyncratic way, the value locked up inside individuals and lost the moment they leave the room. The Industrial Revolution did not make artisans faster; it reorganized their craft into flow, division of labor, and legible process — and unlocked a century of compounding value. Knowledge work is still waiting for that transition. AI is the tool that finally makes it possible: not a bigger workshop, but the first real factory for thought.
And here the two tracks become one story. They were never truly separate. The century-long climb of our tools from noise to lossless — the first track — is exactly what hands us the instruments to make our work lossless — the second. The same pursuit of Lossless Communication that spun out the transistor, the operating system, the network, and now AI was, all along, building the equipment for a factory it had not yet imagined. A hundred years of squeezing value out of the wire produced the machinery; only now do we turn that machinery back on the knowledge work itself. Two arcs, one arc. They met in our moment.

Implications

If the arc is real, it tells us where to stand. The frontier is always the same: find the technology that is still lossy — where value is still trapped and access is still narrow — and push it toward lossless, in the hands of as many people as possible.
That is the work I care about now. The most lossless tools we have — durable data storage, website creation, email marketing, and the rest of the modern stack — are still, for most people, lossy in practice: too expensive, too complex, or locked behind gatekeepers. The task is to make sure more and more people can reach them, the way the camera reached everyone and the telephone reached every home.
More to come on what this means for how we work — and for working in the open, portable formats that keep value from evaporating: Markdown, HTML, CSS, and JSON.

Footnotes