David Shapiro · Tech & AI
TIER 5 Wed, 14 Jan 2026 16:37:18 +0000
How many people will the economy support after AI and robots are better, faster, cheaper, and safer than humans at all economically valuable tasks?
͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏
| |
---|---|---
| | | Forwarded this email? Subscribe here for more
You read that right. It was not a typo.
According to my models, only 15% of working age adults will be employed once AI and robotics take off.
The only assumption you need to make to see the writing on the wall is this:
Machines (AI and robots) will be better, faster, cheaper, and safer than humans at all economically valuable tasks.
You can call this Supply Side Saturation.
In other words, "production will be fully automated" and therefore the picture switches to demand-side.
The only question that matters is "in what conditions will humans pay a premium for the privilege of human input?"
A statutory job, or de jure role, is one that is required by law. In some cases, it's simply because the regulations never contemplated that a machine could replace a human, like doctors and lawyers.
But AI and robotics are already proving that licensed knowledge work is no real barrier.
Another type of statutory role is based on accountability or liability. We're talking judges, politicians, and corporate officers (like CEOs and chairmen) because, well, you gotta be able to pin the blame on a real person.
Liability attaches to a person, not a machine.
Licensed Professionals: Doctors, lawyers, engineers
Government Officials: Sheriffs, politicians, judges
Corporate Officers: CEOs, chairmen, boardmembers
All told, these jobs might stay the same into the future. They might also contract or expand.
For example, I could easily foresee a future where doctors and lawyers go the way of the dinosaur. Why would you go to a human doctor when a robot doctor is better, faster, cheaper, and safer than a human?
Consider that
Utah is now allowing AI to write prescriptions
. Doctors, you are officially on notice.
Conversely, we might end up with more public servants as there are more job seekers. Making government more service-oriented by having more employees. We also might have more (or fewer) corporate officers.
In total, my models show 1.5% to maybe 2.5% of LFPR (total labor force participation rate) which, in a US population with 260 million working age adults, translates to 3.9 million to 6.5 million statutory jobs in a post-labor world.
The biggest barrier to growth here is the fixed amount of attention that humans have to offer. It's pretty easy to calculate.
As of right now, America has about 340 million people.
If we assume that work goes away for most, and we optimistically estimate that people will, on average, have 8 discretionary hours per day (other than sleeping, cleaning, eating, etc) that totals 2.7 billion hours per day that can get monetized somehow.
It sounds like a lot, until you run the numbers, and realize that everyone will be competing over that finite bucket.
It becomes even more depressing when you realize that the attention economy is drastically unequal. The top 1% of creators take 30% to 50% of all attention. The vast majority of content creators already do not make a living on the attention economy, and those numbers will only get worse as more people flood into the market.
However, not all is lost.
The attention economy is not just on the global internet. Creators are in the "one-to-many" bucket, essentially the new broadcast media.
There are two other buckets.
One-to-few bucket (cohorts of 5 or more)
One-to-one bucket (individualized attention)
Even so, all of these buckets will be competing with Netflix and Amazon for time and attention, as well as content creators, personal coaches, therapists, and so on.
To make matters worse, is that the winners of the attention economy ALSO are in high demand for one-to-few cohort classes and one-on-one coaching.
Any way you slice it, the attention economy cannot possibly absorb tens of millions of displaced workers. Not in America, not anywhere.
All told, you're looking at 10% to maybe 15% (on a good day) of the population will be employed in some attention/care/experience sector job.
Celebrities and influencers
Communities organizers and social infrastructure builders
Personal coaches, fitness, spiritual, emotional, and relational help
Care providers for elderly and children
But again, don't bank on some of these, because once robots are good enough, it will still be cheaper to have C3P0 keeping grandma company than a full-time live-in aid.
"But Dave, technology always creates new jobs!"
No it doesn't. It never has.
Technology has created new goods and services, true. The iPhone did not exist 20 years ago. Netflix streaming did not exist 20 years ago.
The demand for those goods and services happened to accidentally create new jobs.
But there is no law of physics or economics that says "new goods and services in the marketplace must be provisioned by human input."
It just so happens that, until now, human input was always necessary.
Remember, the consumer based economy is driven by consumer demand for goods and services. There's nothing in macroeconomic theory that says "consumer economies necessarily require labor and jobs."
The only thing the economy needs to thrive is individuals with money to spend. It does not matter how they got the money.
"But Dave, there's a functionally infinite amount of work to be done! Cancer research, space exploration, and all that!" Which is true.
But again, there's no economically rational reason that a human brain and body must be involved. In fact, human involvement might become negative EV (expected value) because humans are slow and error-prone.
This is what it means to solve supply-side, or functional productivity.
So what about demand side? How do we shore up demand?
The only thing that matters to a consumption driven economy is household ability to purchase. In market economies, household consumption constitutes 70% to 85% of GDP.
Right now, wage-labor allocation is the primary way that wealth is distributed.
However, we are already " demand constrained" meaning that GDP growth is hamstrung by low household spending, and household spending is constrained (you guessed it!) low wages.
We are already caught in the gravitational pull of a techno-deflationary death spiral.
Post-labor economics began decades ago, and all the economic pain we're feeling today? That's just the symptoms of the fever taking over.
So, how do we fix this?
Household income arrives from three buckets:
Wages(labor, salaries, etc)
Capital(business ownership, stocks, bonds, rentals, etc)
Transfers(government payments like welfare, food assistance, subsidized healthcare, rental assistance)
That's it. Those are the three sources of household income. Those are the three primary catalyzers of aggregate demand.
Since wage labor is going away, that leaves two options to shore up household demand: capital based income and government transfers.
Well, we don't want to be dependent upon a welfare state. I do believe that some UBI and redistribution and predistribution is necessary to create a floor.
Either way, the calculus is braindead simple: We must increase capital participation to shore up household income.
It's as simple as that.
But what does that look like? There are dozens of options.
Universal Basic Capital(or capital endowments, like baby bonds)
Sovereign Wealth Funds(dividends paid by collective ownership)
Mandatory Capital Disbursements(require companies to sell shares)
ESOP/EOT(employee owned companies)
Cooperatives and Collectives(productive assets and businesses that are collectively owned)
These are just a few categories off the top of my head.
That may sound complicated so let me break it down for you. In this Post-Labor Economics future, where are you getting money from?
Checks from your local or municipal government. Urban wealth funds will pay out on a quarterly or annual basis. These are monetized by fund raisers, property taxes, and other productive activities. You can contribute directly to these through community enrichment programs, such as shared labor.
Checks from your state or provincial government. We already see examples such as Alaska's Permanent Fund and New Mexico's land trust fund. This is the lowest level that a sovereign wealth fund really makes sense, though urban wealth funds can be capitalized the same way.
Checks from your federal or national government. This includes sovereign wealth funds, dividends from collectively owned assets, capitalized endowments, and so on. Norway already has the largest example of this globally.
Now, you might say "Dave, this is all still coming from the government, so is this not a transfer?"
No, it is not.
A transfer is where the government taxes the rich and gives to the power.
An endowment or wealth fund is when the government runs an annuity for everyone. You get paid based on dividends and other proceeds from ordinary investments.
This creates a growing floor for everyone.
But that's only step one.
If we can create a multi-layered "UBI" based on capitalized sovereign wealth funds, we should also create more private on-ramps to capital accumulation.
Plenty of jurisdictions and polities already have things like mandatory savings and retirement accounts, but that's entry level stuff.
There is more that we can do to incentivize, reward, and encourage capital accumulation at the individual level.
Tax breaks for all capital gains for low net-worth individuals.
More investment vehicles such as DAOs, ESOPs, coops, and so on.
Negative income taxes and dollar-matching investment schemes.
Tax-advantaged accounts and tax breaks on qualified investments.
In other words, we make it stupidly simple to invest and accumulate wealth, with your floor being provided by collective wealth.
Thus, you become a venture capitalist. You become a hedge fund manager. That's your primary "job" in this future.
The biggest complaint I get when I talk about this stuff is "the elites will never go for this."
False.
It is in their best interest to support these policies. And I don't mean "because we'll grab our pitchforks and introduce them to Madame Guillotine."
It's far simpler than that, and it has to do with greed. Rational self-interest. Plain and simple.
Here's how it works:
In order to make money, elites need paying customers. Right now, their customers are broke, and losing purchasing power by the day.
Velocity of currency is too low. You can't build a company if there's no one around to spend it money.
Remember, the vast majority of GDP is driven by household spending. So, why don't we start implementing policies to ensure that households have money to spend?
The government wins. The elites win. Voters win.
It seems obvious to me (granted, I've been knee-deep in this stuff for the last 2 years straight!).
Any way you slice it, if you make only one assumption: that machines will be better, faster, cheaper, and safer than humans at any economically valuable task, then mass unemployment is the inevitable future.
But this is not a bad thing.
This is liberation.
Labor Zero is about liberating humanity from the drudgery of wage labor. It's about shattering the institution of wage slavery.
And we replace it a simple new distribution mechanism: capital.
Wages -> Capital.
It's that simple.
The central question--how many humans will work in a fully automated economy--reduces to a simple identity:
LFPR = Total Worker-Hours Demanded ÷ Hours Worked per Participant
Where total worker-hours demanded equals the sum of two components: (1) hours consumed in human-premium services, adjusted for scalability, and (2) statutory/legitimacy labor requirements.
Expressed algebraically:
Worker-hours per capita per year = 52 × (h_week / s) + h_stat
Where:
h_week = weekly hours of human-delivered service consumed per person
s = scalability factor (consumers served per worker-hour)
h_stat = statutory labor hours required per capita annually
The LFPR then equals worker-hours divided by average hours worked per participant (H_part). The critical insight: even if total labor demand is only 10% of current levels measured in hours, LFPR can reach 15-20% if average hours worked drops to part-time levels.
Human attention is the ultimate binding constraint. Unlike every other economic input, attention cannot be expanded by technology.
Global attention budget:
World population: 8 billion
Allocable waking hours per day: ~14 hours (after sleep, eating, hygiene, basic maintenance)
Daily attention budget: 112 billion person-hours
Annual attention budget: ~41 trillion person-hours
This is the entire pie. Every job that depends on human consumption--which is nearly all remaining jobs in the automated equilibrium--must fit within this fixed budget.
US attention budget:
US population: ~340 million
Discretionary hours per day: ~8 hours (conservative estimate for pure leisure/consumption time)
Daily discretionary attention: 2.7 billion person-hours
Annual discretionary attention: ~986 billion person-hours
No technology, preference shift, or policy intervention can expand these numbers. The 24-hour day is immune to Jevons paradox.
Attention Architecture and Power Laws
Different job categories have radically different relationships to the attention constraint:
One-to-Many (Broadcast):
Leverage ratio: 1 producer serves 10,000 to 100,000,000 consumers
Power law concentration: Extreme. Top 1% capture 50%+ of attention; top 0.1% capture ~30%
Examples: Content creators, celebrities, professional athletes, thought leaders
Employment capacity: Very low. Perhaps 2-5 million viable positions globally; 200,000-500,000 in the US
Winner-take-all dynamics make this category brutal for most participants
One-to-Few (Cohort):
Leverage ratio: 1 producer serves 5-50 consumers per session
Power law concentration: Moderate. Quality matters but local markets fragment competition
Examples: Teachers, workshop leaders, tour guides, small venue performers, group fitness instructors, restaurant service
Employment capacity: Medium. 8-15 million positions in the US
Geographic and relational fragmentation defeats extreme concentration
One-to-One (Relational):
Leverage ratio: 1:1 (no leverage--provider hours equal consumer hours)
Power law concentration: Weak. Trust and relationship quality matter more than being "best in world"
Examples: Therapists, coaches, caregivers, personal trainers, artisans working on commission
Employment capacity: High. 10-25 million positions in the US
Markets naturally fragment into local, trust-based networks
Many-to-One (Governance):
Leverage ratio: Inverse--many people's activities flow through one decision-maker
Power law concentration: Institutionally constrained by design
Examples: Judges, corporate officers, elected officials, regulators
Employment capacity: Fixed by institutional architecture. 4-6 million positions in the US
The key policy implication: favoring one-to-one over one-to-many architectures increases total employment and reduces inequality.
Statutory jobs are positions where legal frameworks require human accountability, decision-making authority, or physical presence.
Corporate Governance:
Approximately 6 million business entities in the US
Average of 1.5 officers per entity (accounting for individuals serving multiple boards)
Estimated unique individuals: 2-3 million
Licensed Professionals with Mandatory Human Involvement:
Physicians (current: 1.1 million; persisting for signing authority: 200,000-400,000)
Attorneys (current: 1.3 million; persisting for court/filings: 200,000-400,000)
Judges and magistrates (current: ~34,000; persisting: 30,000-50,000)
Notaries and attestation roles: 50,000-100,000
Elected Officials and Political Apparatus:
Elected officials at all levels: ~500,000
Required political staff and advisors: ~500,000
Total: ~1 million
Regulatory and Fiduciary Roles:
Financial fiduciaries where human responsibility is mandated
Regulatory officials and inspectors requiring human sign-off
Estimate: 500,000-1 million
Statutory Total: 4-6 million workers, representing 1.5-2.3% of working-age population
Note: This category could expand significantly if regulation mandates human-in-the-loop oversight of AI systems, or contract if professional licensing requirements are relaxed.
One-to-Many (Content/Celebrity):
Consumer attention available for broadcast content: If average person spends 4 hours/day on one-to-many consumption, that's 11.7 trillion viewer-hours annually (global) or approximately 500 billion viewer-hours (US).
A mid-tier successful creator captures roughly 500,000 viewer-hours per year. Naive division suggests 23 million "creator slots" globally.
Power law adjustment devastates this number:
Top 0.1% capture ~30% of attention
Top 1% capture ~50%
Top 10% capture ~80%
Bottom 50% capture ~5%
Viable creator positions (sufficient attention capture for meaningful employment): 2-5 million globally, 200,000-500,000 in the US.
Breakdown:
Professional athletes (top tier): ~50,000
Celebrity entertainers: ~100,000
Viable content creators: 100,000-300,000
One-to-Few (Experiences/Education):
Consumer attention available: If average person spends 2 hours/day in facilitated group experiences, that's 5.84 trillion consumer-hours annually (global).
Calculation: Average group size of 12 persons, provider works 1,500 hours/year. Required providers: 5.84T ÷ 12 ÷ 1,500 = 325 million globally (theoretical maximum)
Market clearing and participation adjustments reduce this to 100-200 million globally, with US share of 8-15 million.
Breakdown:
Teachers/instructors (preference-based): 2-4 million
Hospitality (restaurants, events): 3-5 million
Tour guides, facilitators: 500,000-1 million
Small venue performers: 300,000-500,000
Group fitness, wellness: 500,000-1 million
Workshop leaders, craft instructors: 500,000-1 million
One-to-One (Relational Services):
Consumer attention available: If average person spends 1.5 hours/day receiving one-to-one human services, that's 4.38 trillion consumer-hours annually (global).
One-to-one means no leverage: provider hours equal consumer hours. At 1,500 provider hours/year, the theoretical maximum is 2.9 billion providers globally--obviously absurd.
Realistic participation (5-10% of population willing/able to provide at meaningful level): 400-800 million globally, with US share of 10-25 million FTE.
Breakdown:
Therapy, counseling, coaching: 1-2 million
Personal training, wellness: 500,000-1 million
Eldercare (human preference): 2-4 million
Childcare (human preference): 1-2 million
Personal chefs, household staff: 500,000-1 million
Artisanal craft (commission-based): 1-2 million
Healthcare companions: 1-2 million
Social infrastructure, community facilitation: 1-3 million
Three scenarios capture the plausible range:
Scenario 1: Thin Human Premium (Floor)
Assumptions:
h_week = 2 hours (human services are rare luxury)
s = 3.0 (most services are group/broadcast)
h_stat = 15 hours/year (minimal statutory requirements)
H_part = 700 hours/year (work is occasional)
Results:
Worker-hours per capita: 49.7 hours/year
FTE share: 3.3%
LFPR: 7.1%
Employed (US): ~18 million
Scenario 2: Mixed Experience Economy (Central Estimate)
Assumptions:
h_week = 5 hours (human services are normal part of life)
s = 2.0 (mix of 1:1, small group, and broadcast)
h_stat = 25 hours/year (moderate statutory requirements)
H_part = 800 hours/year (work is typically part-time)
Results:
Worker-hours per capita: 155 hours/year
FTE share: 10.3%
LFPR: 19.4%
Employed (US): ~50 million
Scenario 3: High-Touch Culture (Ceiling)
Assumptions:
h_week = 8 hours (human services are cultural default)
s = 1.5 (most services are 1:1 or small group)
h_stat = 30 hours/year (expanded statutory requirements)
H_part = 900 hours/year (work is regular but not full-time)
Results:
Worker-hours per capita: 307.3 hours/year
FTE share: 20.5%
LFPR: 34.1%
Employed (US): ~89 million
Consolidated Central Estimate
Aggregating across categories with attention constraints:
By Category (US, millions):
Statutory/Governance: 4-6M
One-to-Many (Content/Celebrity): 0.2-0.5M
One-to-Few (Experiences): 8-15M
One-to-One (Relational): 10-25M
Total: 22-47M
Against US working-age population of 260 million:
Floor LFPR: 5-6% (only statutory + core authenticity)
Central LFPR: 12-15% (robust demand preferences + attention constraint)
Ceiling LFPR: 18-20% (strong preferences + expanded statutory + work-sharing)
Current US LFPR is approximately 62% with ~161 million employed. The central estimate represents a 75-80% decline from current levels.
The saturation dynamic emerges from zero-barrier entry combined with fixed demand.
Current labor markets are segmented by capability barriers: credentials, talent, training, connections. These barriers limit competition and support wages.
In the automated equilibrium, barriers dissolve for most remaining work categories. Content creation requires no credential. Coaching certification can be obtained. Artisanal skills can be learned. Care work requires empathy, not scarce credentials.
The math of saturation:
Working-age population: 260 million
Percentage who would like to participate in meaningful work: ~30% (conservative)
People seeking work: ~78 million
Attention-constrained positions available: 30-40 million
Structural excess supply: 40-50 million people who want work but cannot find adequate demand
This is qualitatively different from current unemployment:
Not frictional (not temporary mismatch)
Not cyclical (cannot stimulate attention into existence)
Not structural (training doesn't help when everyone can do these jobs)
Every niche becomes saturated. Content creation, coaching, artisanal craft, local services--all face a flood of supply competing for finite attention.
Baumol 's Cost Disease (Modified):
Classic formulation: Labor-intensive services get relatively more expensive as productivity rises elsewhere (orchestras can't get more productive, so musicians get relatively more expensive).
In the automated equilibrium: Human labor IS the Baumol sector. Everything that can be automated gets cheaper; human-delivered services get relatively more expensive.
Twist: This sustains employment rather than undermining it. The relative expensiveness of human labor maintains the wage premium that makes human work viable. If human labor got cheap through competition, fewer people would bother providing it.
Jevons Paradox (Does Not Apply):
Classic formulation: Efficiency gains increase total consumption of a resource (cheaper coal -> more coal burned; cheaper compute -> more compute used).
Does NOT apply to attention: You cannot consume more attention by making attention capture more efficient. The 24-hour day is immune to Jevons.
Implication: Attention remains absolutely scarce even as everything else becomes abundant.
Sensitivity Analysis
The model is most sensitive to four parameters:
1. Consumer Hours of Human Service (h_week)
This is the largest driver of variance. Range: 2-8 hours/week.
At 2 hrs/week: LFPR of 5-8%
At 5 hrs/week: LFPR of 12-18%
At 8 hrs/week: LFPR of 25-35%
2. Scalability Factor (s)
The weighted average of consumers served per provider-hour. Range: 1.5-3.0.
At s=3.0 (mostly group/broadcast): Lower LFPR
At s=2.0 (mixed): Central estimate
At s=1.5 (mostly 1:1): Higher LFPR
3. Hours Per Participant (H_part)
Average annual hours worked by those in the labor force. Range: 700-1200 hours/year.
At 1200 hrs/yr (near full-time): Lower LFPR for same total hours
At 800 hrs/yr (part-time norm): Central estimate
At 500 hrs/yr (occasional/gig): Higher LFPR for same total hours
4. Statutory Thickness (h_stat)
How much human labor is legally required per capita. Range: 15-60 hours/year.
At 15 hrs/yr (minimal): ~1% LFPR contribution
At 25 hrs/yr (moderate): ~2% LFPR contribution
At 60 hrs/yr (extensive mandates): ~3-4% LFPR contribution
Even within the 12-15% central estimate, distribution is highly unequal.
Composition of remaining employment:
Statutory/Governance: 15-20% of jobs
One-to-Many: 1-2% of jobs
One-to-Few: 25-35% of jobs
One-to-One: 45-55% of jobs
Income distribution:
Statutory roles: Stable, high wages
One-to-Many: Extreme winner-take-all (top 1% earn majority of category income)
One-to-Few: Competitive, moderate inequality
One-to-One: Fragmented, local wage variation
The aggregate LFPR might be 15%, but "good jobs" (stable, well-compensated) might be only 5-8%. The remainder are precarious, part-time, or hobby-level participation.
For comparison:
US LFPR before women entered workforce en masse (~1950): ~59%
US LFPR peak (~2000): ~67%
US LFPR current: ~62%
Projected LFPR under full automation: 12-15%
The decline from 62% to 15% represents a larger shift than the entire increase from women's workforce participation.
Model Validation
Two independent modeling approaches converge on similar estimates:
Approach 1: Category-based bottom-up
Sum employment across statutory, one-to-many, one-to-few, and one-to-one categories, adjusted for overlap. Result: 22-47 million employed, or 8-18% LFPR.
Approach 2: Attention-hours top-down
Calculate total human attention available, divide by scalability-weighted service delivery, add statutory requirements. Result: 10-20% LFPR depending on parameter assumptions.
The convergence of independent approaches on similar ranges (central estimate: 12-15%) suggests structural validity rather than methodological artifact.
Upside risks (could increase LFPR):
Stronger-than-expected authenticity preferences
Expanded statutory human-in-the-loop requirements
Cultural attachment to work-as-identity proving durable
Work-sharing norms developing organically
New categories of human-valued activity emerging
Downside risks (could decrease LFPR):
Authenticity preferences proving weaker than expected
Statutory requirements being streamlined away
Care preferences shifting toward robot provision
Power law concentration intensifying across categories
Social acceptance of non-work identity accelerating
Under the single assumption that machines become better, faster, cheaper, and safer than humans at all economically valuable tasks:
Human labor demand becomes bounded by attention, not productivity
The attention constraint caps LFPR at approximately 20% regardless of preferences
The defensible central estimate is 12-15% LFPR
This represents a 75-80% decline from current employment levels
The remaining employment concentrates in relational services (largest), experience provision (second), statutory roles (stable), and celebrity/content (smallest)
Distribution within this smaller labor force becomes the critical social question
The numbers are stark, but they follow necessarily from the stipulated premises. The only question is whether the premises are correct and how quickly they arrive.
You're currently a free subscriber to David Shapiro's Substack. For the full experience, upgrade your subscription.
Upgrade to paid
(C) 2026 David Shapiro
548 Market Street PMB 72296, San Francisco, CA 94104
Unsubscribe