That is the rough idea behind Human Reserved, Bill Gates’s recent proposal to preserve certain roles for people even when artificial intelligence could perform them. His concern is prospective: how do we keep working people from being displaced and discarded?
My mind went to people our labour market has already discarded—or never found a way to receive.
For years, I’ve worked around homelessness and observed one of the most human forms of labour: being present and decent with someone lying on a sidewalk, living in a shelter, or struggling through addiction.
The need is everywhere, but the work is strangely scarce. Neighbours may avert their eyes. Transit employees, hospital staff, and other professionals encounter people in crisis but may lack the time, role, training, or trust required to reach them. Work that everyone encounters can become work that nobody owns.
Addiction can be difficult to comprehend without having lived through it. Someone who has survived it may recognize its rhythms, evasions, and possibilities differently.
Gates asks which jobs we should reserve from machines. I wonder which paths into work we should reserve for people whom ordinary hiring has already screened out.
Could we build supported job on-ramps—sometimes beginning with only a few accountable hours—for people able to understand and reach others whom our institutions find hardest to engage?
Hamilton offers one answer.
In a September 1 Globe and Mail report, Molly Hayes follows Madison Tatlock and Matthew Cheeseman through the halls of St. Joseph’s Healthcare Hamilton. Three and a half years earlier, they had occupied those halls as patients in active addiction. Both had been homeless. Their baby was placed in foster care. Neither had worked in years, and both had criminal records.
Now they travel the same halls wearing hospital identification badges.
Tatlock and Cheeseman are paid members of a hospital-wide peer-support program for patients struggling with substance use. Those patients may have entered hospital for a heart attack, surgery, or another condition. But withdrawal, cravings, and distrust can interfere with their care—or cause them to leave before it is complete.
A peer worker may interrupt that sequence.
Sometimes the work begins with entering a room and saying: I have been there too. Peers take patients outside for cigarettes, help replace identification, connect people with services, and sometimes help them preserve their housing. The small team has logged more than 14,000 patient interactions since March 2025.
People who have navigated addiction, homelessness, recovery, and hospital care know this unusually difficult problem from the inside. Their experience can give them credibility with patients whom the institution struggles to reach. Hospital staff now seek their advice.
Credibility can produce trust. Trust may keep someone connected to care long enough for other help to work.
The peers also model a possible future. Tatlock does not merely tell another patient that change is possible. She walks into the room as evidence: someone once desperately unwell in the same hospital now returns as a colleague, parent, and paid professional.
Experience that carried stigma becomes expertise. A person who needed help becomes someone able to give it.
The couple’s recovery began before this employment. The job did not rescue them. It gave recovery somewhere to go: into responsibility, income, purpose, and further learning. Tatlock has since enrolled part-time in a social-service-worker program.
Capability begins producing more capability.
None of this happens automatically. Lived experience is not by itself a qualification, and suffering should not become one. St. Joseph’s created paid positions within a team that includes physicians, social workers, and nursing support. Selection, training, responsibility, and institutional backing turned potential into accountable work.
Among the hundreds of people I know and continue to meet in and around shelters, I repeatedly encounter knowledge, energy, and social skill that ordinary job descriptions are not organized to receive. What is often missing is a supported job on-ramp for people who have become disconnected from ordinary employment: easier to enter, but leading to serious work.
Here, then, are two ambitions. Gates proposes a softer landing before displacement begins a downward spiral. Hamilton offers a possible path upward after exclusion has already compounded. After chronic use of emergency resources has already spiraled.
By choosing a path of growth, unusually wide spillovers may result. A worker gains income, rhythm, skills, and standing. A family gains stability. A patient encounters someone credible and may remain in care. A hospital may avoid another premature departure and its spillover, a more costly return. Recovery becomes visible to someone who cannot yet imagine it.
One path back becomes evidence of another.
The pilot has not proved every link in that chain, and its future funding remains uncertain. But it reveals a positive mechanism inside a field usually described through spirals of crisis, cost, and decline.
People widely treated as residual may possess useful capital that our institutions have not learned to recognize, develop, or employ.
Gates proposes work reserved for humans. Hamilton has begun building a human way back to work.
Where else might institutions create supported first steps into useful employment, or even engagement?
And which forms of work might do more than prevent a spiral; which might help reverse one?
Molly Hayes, “For Hamiltonians battling addiction, peer support is a dose of empathy,” photographs by Laura Proctor, The Globe and Mail, September 1, 2026.
I began complaining about misconduct in finance in 1990. Across four decades, the record of matters I raised has come to include permanent bans, collapsed firms, billions in damage, and settlements measured in hundreds of millions.
The amounts are alarming. I am more interested in the time.
This is an essay about institutional learning: what organizations learn when they repeatedly fail to respond, what executives learn when warnings bring no visible consequence, and what victims learn when speaking up changes nothing.
A warning takes work. Someone notices something wrong, checks it, writes it down, finds the right door, and knocks. Often, the person must knock again: another letter, another call, another explanation to someone encountering the problem for the first time.
Sometimes the work costs much more than time. In two of the cases below, I was directly involved, not merely observing. In each, becoming a whistleblower meant relinquishing more than $1 million in carried interest.
Those choices can damage careers. Lawsuits, lost income, professional isolation, and the prospect of being labelled difficult all chill people who possess inside knowledge. Chilling them does not help institutions learn.
Then nothing changes.
The conduct appears to remain non-urgent precisely because the institution is already getting away with it.
Meanwhile, the people whose conduct prompted the warning are also observing what happens. A practice survives one complaint, so its apparent risk falls. It survives another, and confidence rises. Conduct persists; then it hardens. Executives win.
On one side, an institution learns not to hear. On the other, those benefiting from delay learn not to fear.
The resulting advantage is what I think of as the return on delay. Benefits arrive now. Correction remains uncertain and distant. Each additional year of survival makes the conduct look more normal, more established, and perhaps more defensible.
Delay can function like financing. The people harmed advance the capital; any correction comes years later.
Sometimes nothing happens because responsibility is fragmented or the warning reaches someone without authority. In harder cases, listening threatens revenue, status, or careers. Silence may be useful.
I have begun calling one side of this process learned deafness: an institution becomes progressively less able to receive information that is inconvenient, uncomfortable, or unfamiliar.
Meanwhile, the information reaching it becomes distorted. People who see that nothing changes may stop knocking. Fewer warnings then arrive. Executives can mistake that silence for evidence that the problem is small—or that things are improving.
That is how learned deafness compounds: inattention discourages warnings; fewer warnings produce false reassurance; false reassurance permits greater inattention.
Five warnings
These are not five randomly selected scandals. They are five matters I encountered personally—five occasions on which I saw something, raised a concern, and then watched time add information.
Noram
In 1994, Noram advertised a 35.2 per cent “average annual return” for its standard leveraged account. On the same page, it called itself “financially conservative” and said it minimized risk by investing exclusively in high-grade bonds. I specialized in investment-performance measurement. I could not reconcile those claims. So I called the newspaper’s advertising department, Noram’s clearing broker, and the OSC—repeatedly.1
Advertisements like this ran for at least five years.
Then the OSC suspended Noram. It later permanently barred Andrew Willman and found that the firm had made misleading or ambiguous representations about risk and conservative investing.2
A 1994 Globe and Mail advertisement for Noram Capital Management claiming a 35.2 per cent average annual return for a standard leveraged managed account while describing the approach as financially conservative and invested exclusively in high-grade bonds. Advertisements carrying versions of these performance claims appeared for at least five years. In 1999, the OSC suspended Noram; in 2001, Andrew Willman was permanently barred from trading securities.
Trailer commissions
Throughout the 1990s, I complained about mutual-fund trailer commissions paid to discount brokers. The contradiction was plain. Trailer commissions were intended to compensate dealers for continuing advice. Discount brokers were prohibited from providing it. Yet the commissions continued to be extracted from investors’ funds.
Canada’s banks built, bought, and expanded these discount brokerages. Each year of bank ownership and industry-wide acceptance made the contradiction appear more normal, more lucrative, and harder to reverse.
Decades later came the class actions. Announced settlements involving funds managed by CIBC, RBC, and TD now exceed $140 million. Other proceedings continue. That is a great deal of eventual correction. It does not answer the underlying question: What did the intervening decades pay?3
AIC and market timing
More than twenty years ago, I began hearing rumours about a small trading team in a small town moving tens of millions of dollars into and out of retail mutual funds—often.
The practice was called market timing. Traders exploited stale prices in funds holding overseas securities, entering before those prices caught up and leaving after they did. Frequent traders profited at the expense of long-term investors.
I complained while it was happening.
The scale of the eventual reckoning was enormous. In 2004, four fund managers agreed to return money to affected investors: Investors Group, $19 million; AGF, $29 million; CI, $49 million; and AIC, $59 million. The following year, Franklin Templeton agreed to return another $49 million.
Together: more than $200 million.
The AIC disclosure also resolved an old puzzle for me. In September 1999, I had warned in The Globe and Mail that redemptions from AIC’s enormous, illiquid holdings could force sales, depress its daily price, and provoke still more redemptions. I later learned that three large market timers had been allowed to move rapidly in and out of AIC funds. Their repeat investments supplied liquidity I could not see when I wrote the warning.
A problem can remain hidden while an accommodation keeps it running.
Still, correction had not finished travelling. In 2026—more than a quarter-century after the conduct began—an Ontario court ordered AIC and CI to pay a further $170 million in damages and interest. CI said it expected to appeal.
Portus
On February 26, 2004, I warned Portus’s principals that they were surrounded by yes-men and missing critical warnings. The next day, I began trying to alert the OSC. My call records show thirteen voicemails. I also submitted key documents. No call came back.
Eleven months later, the OSC barred Portus from accepting new money. Thirteen months after my warning, the firm entered receivership.
Other critics faced more than silence. Portus sued people who questioned its representations. The threat of blowback chilled competitors who possessed useful information.
Regulatory intervention did not end the delay. It took nearly five more years, extensive recovery work, and a $612-million settlement with Société Générale before investors recovered roughly 93 cents on the dollar. Manulife had already made its own clients whole by assuming approximately $245 million of their Portus investments.⁶
Jon Chevreau later examined the warning I had sent nearly a year earlier and wrote: “Young’s missive, read today, seems nothing less than prophecy.” 7
It was not prophecy. It was information available before the danger acquired institutional force.
A current test
The fifth matter remains unresolved. It is a decade-long complaint involving the insurance industry. I began warning the board more than four years ago. The executive suite has since changed. By my calculation, shareholders have lost roughly $3 billion in market value during the period. I believe the larger consequence has yet to arrive.
I have told authorities that the evidence raises another question: Did insiders or counterparties silently get risk off while customers and ordinary shareholders remained exposed?
I think my testimony on the phishing part of this matter warrants criminal scrutiny.
After warning the executives, the board, the auditor, and several regulators, must I take the evidence to the police myself?
What the warnings taught
These five encounters are not a representative sample. They cannot tell us how often institutions fail to hear.
No institution can investigate every assertion. Complaints may be mistaken, incomplete, or impossible to prove. But an institution must remain capable of receiving difficult information.
If one analyst nevertheless encountered this pattern repeatedly across four decades, the mechanism deserves investigation.
How many warnings disappear before anyone begins counting them?
The usual record begins late. It tells us when a regulator opened a proceeding, when a board announced an investigation, or when a court approved a settlement. It rarely captures the earlier period in which customers, employees, competitors, or specialists were already trying to make the problem heard.
Those missing warnings matter. They could tell us what information was available, who received it, and what happened next.
In the harder cases, silence may be cultivated. Critics encounter delay, legal threats, professional blowback, or the wearying demand to explain the entire problem again.
The warning does not have to be disproved. Its bearer merely has to be exhausted—or taught to fear what speaking again might cost.
The people trying to warn an institution learn which doors never open, how much exertion another attempt will require, and eventually that nothing changes. Their withdrawal then distorts the institution’s view of itself. Fewer complaints arrive. Management sees less opposition. Silence begins to look like satisfaction.
An institution can teach people not to trust it.
This is why I have begun thinking of a Hard Case Method: start with consequential cases in which early warnings can be recovered, then compare what was knowable with what institutions did. Trace the warning through the system. Find where it stopped, what incentives surrounded that point, and how the delay changed behaviour.
The purpose is not to prove that every complainant was right. It is to learn why some warnings receive serious attention while others must wait for catastrophe to become legible.
How can institutions make early attention more valuable than prolonged avoidance—especially when avoidance pays?
And can institutions predict, prevent, and manage the blowback when a preventable loss finally makes years of ignored warnings visible?
Sources and notes
1. Noram advertisement, “Noram Performance Record Speaks for Itself,” The Globe and Mail, February 9, 1994, p. B15, accessed through ProQuest Historical Newspapers.
2. Ontario Securities Commission, Settlement Agreement in the Matter of Noram Capital Management Inc. and Andrew Willman, February 9, 2001. Source
3. Siskinds LLP, “Mutual Fund Trailing Commissions Class Action.” The page reports announced settlements of $26 million with CIBC, $45 million with RBC, and $70.25 million with TD; the RBC settlement was awaiting court approval when consulted. Source
4. Duff Young, “A dimmer view of AIC Advantage,” The Globe and Mail, September 25, 1999, p. B7, accessed through ProQuest Historical Newspapers.
5. Ontario Securities Commission, Settlement Agreement in the Matter of AIC Limited, December 16, 2004; Fischer v. IG Investment Management Ltd., 2026 ONSC 4142. The 2026 award exceeded $170 million in damages and interest and was subject to possible appeal. Source
6. FAIR Canada, A Decade of Financial Scandals; Thornton Grout Finnigan LLP, Portus receivership summary. Receiver’s counsel reports a recovery of 93.13 cents per dollar of claims. Source
7. Jonathan Chevreau, “E-mail warned of problems,” Financial Post, February 16, 2005, p. FP6, accessed through Canadian Newsstream.
The current test describes an ongoing complaint and includes the author’s own calculation and interpretation. The parties are not identified.
Gilbert Gottfried was credited as the original voice of the Aflac duck. But older financial analysts like me know better. It was plainly a recording of Ed Rosenbaum laughing at one of his own jokes.
The laugh was ridiculous. The institution he built around it became the world’s premier CFA study seminar—and remained so for roughly three decades.
I knew the laugh well. Ed had taught me as an undergraduate; I later attended the Windsor Class at all three levels and earned the designation at 24.
Every May, hundreds of worried, overworked investment professionals converged on the University of Windsor. Many travelled across oceans to get there. New York had Wall Street. Boston managed the money. But when it was time to study, they all came to Windsor.
Among generations of analysts, from Sydney to Stuttgart, the seminar became known simply as the Windsor Class.
Its arithmetic presents the first puzzle. A typical seminar brought roughly 600 candidates together with about six faculty—an apparently impossible ratio of around 100:1.
How could one professor help a hundred exhausted adults help one another?
Part of the answer is that the categories were wrong. A professor at the front explained the theory. A candidate at the back, responsible for billions in that particular market, occasionally explained what the theory meant in practice. At Windsor, “student” did not mean beginner.
Nor were the levels sealed off. Rookies studied alongside Level II and Level III candidates; advanced candidates helped people coming behind them. Knowledge travelled down and sideways. So did candid advice: a disinterested, well-informed course correction from someone already doing the work could change a candidate’s preparation—and perhaps a career.
Ambition travelled the other way. Newer candidates saw people with remarkable careers sitting beside them, and returning candidates one or two levels further along the same path. They did not simply learn more. They acquired evidence about who they might become.
Competence was not abstract. It was sitting beside you—older and younger, from every inhabited continent and every corner of finance.
Every May, these giants of finance travelled to Windsor. Some came because they needed help. Others returned because they had help to give. In 1990, one of those alumni volunteers was John Simpson. He had earned his CFA six years earlier and now ran Fidelity Canada.
The seminar also assembled something rarer than expertise: common purpose under real stakes.
Business-school classrooms can reward the student who grandstands or signals that the work is beneath him. Windsor offered little social return for that. These adults had travelled too far and surrendered too much to treat the week casually. The exam for our level loomed heavily over us, all to be written on the same day three weeks hence.
That seriousness changed the room’s social economy. The exam was the adversary, not the person beside you; effort and useful knowledge earned attention. We were all working together not to fail.
The room learned during the week. Ed made the institution learn between years. He did not merely welcome feedback; he demanded it, session by session. The judgments carried consequences: instructors were coached, responsibilities shifted, and the strongest teachers returned. Carl Schweser was among the instructors Ed cultivated; he would carry the work of preparing CFA candidates far beyond Windsor.
Each May, returning candidates encountered a seminar reshaped by what they and their predecessors had told Ed.
Contemporary reporting from the 1990s showed Windsor candidates performing substantially better than the broader candidate population. The results appear to have strengthened Windsor’s reputation, drawing candidates back for the next level while attracting another overwhelmed group of rookies.
Two reinforcing loops had formed. Faculty organized the learning, but candidates helped teach one another. Then candidates evaluated the teaching, helping Ed improve the faculty. Successful candidates became accomplished alumni; some returned as volunteers and placed what they had learned—and what they had since become—back into the room.
About six faculty helped teach 600 candidates. Six hundred candidates helped Ed improve the teaching.
That may explain how the seminar worked. But one detail still nags at me. Canadian mutual funds stood at the beginning of an extraordinary boom in 1990. I could scarcely have imagined a better job in finance than running Fidelity Canada. Yet, for years, John would step away from that job for a week and return to Windsor, simply to help candidates succeed.
What made the Windsor Class worth coming back to?
I have offered one provisional answer: Windsor arranged faculty, candidates, and volunteers around a common purpose. It made effort respectable, moved knowledge across levels, acted on demanding feedback, and left little oxygen for indifference or resistance.
Can other institutions do that deliberately—not by finding unusually virtuous people, but by arranging ordinary roles so that contribution earns status and obstruction finds no audience?
If you attended, taught, or volunteered at the Windsor Class, I would like to know what I have missed. What made it work? What made people return? Which of its arrangements could help adults learn, work, and develop one another elsewhere?
And if people working in one of the world’s most competitive professions could cooperate so effectively in Windsor, what do those of us who benefited owe the communities that helped form us?
