Nobody is starving for information anymore. The world produced 147 zettabytes of it last year. What learners have is indigestion — and the evidence says the cure is not more content but a guide who sells sequence, filter, and accountability.
The learner of 2026 does not have an access problem; they have a digestion problem. Global data grew roughly 73x in fourteen years — 2 zettabytes in 2010 to 147 in 2024 — while the human decision apparatus stayed fixed: under information overload, lab subjects decide 37% slower and decide worse, and a meta-analysis of 31 experiments found more information reduces decision quality. Free self-directed content fails at scale — 52% of 12.67 million MOOC registrants never even entered the courseware. So as content supply explodes, the value of the person who can digest it for you compounds. A mentor in an infinite-content world sells three things no library can: sequence, filter, and accountability.
I file this dispatch against a claim I keep hearing, usually delivered as if it were obvious: "Nobody needs a mentor anymore. Everything is on YouTube." I want to take that claim seriously, because it is half right — and the half it gets wrong is the entire economics of expertise for the next decade. Yes, everything is on YouTube. That is precisely the problem. The learner of 2026 is not starving for information. The learner of 2026 has indigestion.
I did not invent that framing. It is the spine of the field guide I wrote for founders, Indigestion or Starvation, and its opening chapter states the diagnostic in one breath:
"Every founder who has ever scaled a real business has passed through one of two states. Indigestion. Or Starvation. Name yours." — Indigestion or Starvation, Chapter One, "Two States. One Diagnosis."
The book defines the first state precisely: "You have more demand than you can fulfill. The phone rings, the inquiries come in, the work backs up. You are turning revenue away. You cannot deliver what you have already sold." Swap "demand" for "information" and you have described every serious learner I have met in the last three years. The inputs keep arriving. The tabs stay open. The backlog of saved articles, bookmarked threads, and half-watched courses keeps growing. The intake engine works perfectly. The digestion engine does not exist.
Start with the raw numbers, because they are more extreme than most people's intuition. According to SOAX's analysis of Statista's IDC-based Global Datasphere figures, the world created and consumed about 2 zettabytes of data in 2010, 147 zettabytes in 2024 — roughly 402.89 million terabytes every single day — and is projected to hit 181 zettabytes in 2025. That is about a 73x increase in fourteen years, doubling every three to four years. IDC's own forecast, reported by Network World, put the trajectory at 33 zettabytes in 2018 to 175 by 2025 — a 61% compound annual growth rate.
Sources: SOAX research citing Statista/IDC Global Datasphere; IDC forecast via Network World.
You can quibble about which zettabytes are cat videos and which are course material; it does not change the shape of the curve. Content supply in every professional niche — marketing, code, finance, fitness, real estate — is growing at a rate no human intake can track. Meanwhile the thing doing the intake, your brain, ships on the same hardware it shipped on in 2010. That asymmetry is the whole story.
What happens when a fixed-capacity mind meets exponential supply? We do not have to guess; it has been measured, at the level of milliseconds and brain voltage. A 2021 event-related-potentials study in Frontiers in Neuroscience put consumers through online decisions under low and high information load. Under overload, decision time jumped from 1,106.68 ms to 1,511.69 ms — 37% slower, p<0.001. The electrode data explained why: reduced P2 amplitude (fewer attentional resources allocated), smaller P3 (greater decision difficulty), and a larger late positive component — the signature of emotional arousal tied to regret and uncertainty. More input did not make the subjects better informed. It made them slower, more taxed, and more anxious about the choice they eventually made.
And the deficit is not just speed. A meta-analysis by Hwang and Lin in the Journal of Information Science — 31 experiments across 18 empirical studies — found that both information diversity and information repetitiveness have an adverse impact on decision quality. Read that carefully: giving people more varied information hurt their predictions, and giving them more of the same information also hurt their predictions. Past a threshold, there is no flavor of "more" that helps.
The workplace data says we now live past that threshold as a baseline condition. The Microsoft and LinkedIn 2024 Work Trend Index, surveying 31,000 people across 31 countries, found 68% struggle with the pace and volume of work, 46% feel burned out, 85% of emails get read in under 15 seconds, and the average employee spends 60% of their time on communication against 40% actually creating. This is what indigestion looks like at civilizational scale: intake consuming the hours that digestion needed.
If access were the constraint, the MOOC era should have been the greatest democratization of expertise in history. Millions of people, free courses from MIT and Harvard, zero gatekeepers. We ran that experiment for six years, and Reich and Ruipérez-Valiente published the autopsy in Science, "The MOOC pivot". Across 12.67 million registrations in every MIT/Harvard edX course from 2012 to 2018: 52% of registrants never entered the courseware at all. Of those who came back for a second year, retention collapsed from 38% in the first cohort to 7% by 2016–17. Completion rates, in the authors' words, "have not improved over 6 years." And their diagnosis of what was missing is the sentence this whole dispatch hinges on: "human connections through advisers, tutors, and peer groups provide the most important student supports."
This lines up with what instructional science had already established. The landmark review by Kirschner, Sweller and Clark concluded that "minimally guided instruction is less effective and less efficient than instructional approaches that place a strong emphasis on guidance of the student learning process" — with guidance advantageous until learners have enough prior knowledge to guide themselves. Half a century of empirical studies, one direction. Novices given a library flounder. Novices given a guide learn. The library was never the product.
So here is my position, stated plainly: as the cost of content goes to zero, the value of digestion goes up, and digestion is what a mentor sells. Not information — nobody can compete with free on information. A working mentor in 2026 sells three specific things:
Notice that all three are judgment goods, not information goods. They cannot be scraped, and their value rises with content supply, because every new zettabyte makes sequence harder to find, filtering more necessary, and follow-through rarer. This is the same inversion I argued in AI Won't Replace Mentors: the machine multiplies content, which multiplies the market for the person who can digest it. And it is why the working move for an expert is not to publish more but to codify what you know into a sequence someone can actually complete — then price it as the guide-good it is, per the pricing dispatch.
The book's deepest line about founders applies verbatim to learners: "You do not have a discipline problem. You do not have a mindset problem. You have a structure problem" (Indigestion or Starvation, Chapter Two, "Who This Is For"). The learner grinding through their fortieth open tab at midnight does not lack willpower. They lack a structure that meters intake to digestion — and Chapter Three names the only cure that has ever worked for the founder version of the disease: "The answer is always to remove the founder from the middle of the workflow." For the learner, the equivalent move is to stop being the sole curator of your own education. Put a guide in the workflow. Let someone whose judgment you trust decide what enters your queue and in what order.
That is not a retreat from the open internet. It is the only rational response to it. The scroll-everything strategy was optimal in 2010, when there were 2 zettabytes and finding anything was the hard part. At 181 zettabytes, the same strategy is a slow-motion failure mode with lab-verified symptoms: slower decisions, worse decisions, and a completion rate rounding to zero. The market has already noticed. Every "best of" newsletter, every curated cohort, every paid community is the economy repricing digestion. Mentorship is simply the highest-resolution form of that repricing — one expert's digestive system, rented by the person drowning.
Nobody is starving. Everybody is full. The guide who can help you digest is the scarcest asset on the internet, and the numbers above say the scarcity compounds every year.
The Indigestion/Starvation diagnostic is one chapter of a larger system. Indigestion or Starvation: A Field Guide for Founders lays out the full five-step framework — Audit, Separate, Translate, Build, Reclaim — for getting yourself out of the middle of every workflow that does not need you. It runs on the same structural idea as this dispatch, pointed at your business instead of your reading list. Start with the free book here — you just cover $9.95 shipping.
Because the constraint moved. When information was scarce, access was the product. Now that the world produces zettabytes of content a year, the scarce inputs are sequence (what to learn in what order), filtering (what to ignore), and accountability (actually finishing). A mentor sells those three things, not information — and the research on guided versus self-directed learning shows they are exactly what determines whether a learner gets results.
Measurable. A 2021 ERP lab study in Frontiers in Neuroscience found consumers under high information load took significantly longer to decide (1511.69 ms vs 1106.68 ms, p<0.001), allocated fewer attentional resources, and showed higher emotional arousal linked to regret. A 1999 meta-analysis of 31 experiments found that adding either more diverse or more repeated information reduces decision quality. More input measurably produces worse decisions.
They proved the opposite. Reich and Ruipérez-Valiente analyzed 12.67 million registrations across all MIT and Harvard edX courses from 2012 to 2018: 52% of registrants never entered the courseware, second-year retention fell from 38% to 7%, and completion rates did not improve over six years. The authors concluded that human connections — advisers, tutors, peer groups — provide the most important student supports. Free content without a guide mostly produces registrations, not learning.
Three things content cannot deliver: sequence — a curriculum ordered for your specific starting point; filter — permission to ignore 99% of what is published; and accountability — a human who notices when you stall. The half-century of instructional research summarized by Kirschner, Sweller and Clark shows strongly guided learning beats minimally guided learning for novices, which is precisely the mentor’s product.
Author of The Mentor Economy and co-founder of MentorMe. He writes about turning hard-won expertise into AI-leveraged one-person businesses.
The Mentor Economy is the full system — free, you just cover $9.95 shipping.
Get Your Free Copy