Jon's Job Blog

The role I left at PPG was reposted, after I was let go, in Poland. The job ID is a matter of public record, the location is a matter of public record, and the timing — the elimination of my position in the United States and the posting of the substantially similar position in Wrocław — is also a matter of public record, although the company has been careful in its written communications never to describe the two events as a single transaction. They are. Whatever the internal language used to describe what happened, the material outcome was that an analyst function staffed in Pittsburgh was unstaffed in Pittsburgh and re-staffed in Wrocław. The wages associated with the function dropped by approximately seventy percent in the process. The function did not change. The work did not change. What changed was the geography of the person doing the work, and the geography produced a labor cost the company preferred to the labor cost it had been paying.

The transaction described above is so common in white-collar work as to be invisible to most of the people who participate in it. It is the default arrangement of contemporary multinational employment. I do not raise it as a scandal. I raise it because the default arrangement contains, embedded within it, a set of assumptions about the relationship between human dignity and geographic location that the language of global team and distributed workforce has been carefully constructed to obscure. This essay is an attempt to look at the assumptions directly. The argument is that the default arrangement of multinational labor in the present moment is not, as it is typically described, a neutral optimization of resources. It is a set of choices about whose dignity counts where, made by people whose own dignity is rarely the subject of the choices.

The Cost Calculation, Stated Plainly

The economic logic of labor arbitrage is not complicated. A developer in Pittsburgh and a developer in Wrocław with comparable skill, training, and output cost the employing company substantially different amounts. The differential, in the U.S.-to-Poland comparison, is typically in the range of three to four times. A senior analyst whose total compensation in Pittsburgh would run a hundred and twenty thousand dollars annually can be replaced with a senior analyst in Wrocław whose total compensation runs thirty to forty thousand dollars annually. Both analysts are doing approximately the same work for the same employer. Both are subject to the same project deadlines, the same internal politics, and the same expectations of availability. The difference in compensation is not a function of the work. It is a function of geography.

The clearest published version of this arithmetic comes from GitLab, the software company whose unusual practice has been to make its compensation formula public. The GitLab formula applies a location factor to a base salary band, with the factor varying from 1.0 in San Francisco to 0.679 in Wrocław, with other cities slotting in between. The factor is not a measure of skill or contribution. It is a measure of the local cost of labor, which is to say, of how much the company can pay for equivalent work without losing the ability to recruit it. Most companies do this same math. Most companies do not publish the formula. The opacity is the norm; the transparency is the exception. The math, in either case, is the math.

What the math does to the experience of working inside a global team is the subject this essay is actually about. The Pittsburgh analyst and the Wrocław analyst are not abstractions. They are people. They both work on the same projects, attend the same meetings, and produce the same deliverables. The Pittsburgh analyst knows, generally, that the Wrocław counterpart is paid less. The Wrocław analyst knows, generally, that the Pittsburgh counterpart is paid more. Neither one knows the specifics, because the company’s compensation practices keep the specifics private. What both know is that the company is making a decision about how to value each of them that has no relationship to the comparative quality of their work, and that the decision is being made on the basis of where each of them happens to live. The math is on the company’s side. The dignity calculation, on both sides, is somebody else’s problem.

Proximity and Its Decisions

What I observed at PPG, and what I have continued to observe in distributed teams elsewhere, is that proximity becomes the variable that decides a great deal of what is supposed to be decided on other grounds. Proximity, in the management literature, is sometimes presented as a logistical problem to be solved by good tooling — better video conferencing, asynchronous collaboration platforms, written communication standards. The literature is not wrong about the tools. The tools matter. The tools are not the question. The question is what happens when proximity is the variable that, in the absence of conscious effort, decides who is in the room when the room makes the decision.

The person physically closest to the decision-maker has access to the informal conversation, the hallway exchange, the casual sense-check that does not appear on any meeting agenda. The person on the other end of a video link does not. The disparity in informal access is not a malicious choice; it is a feature of how distributed teams have always worked. The disparity becomes a problem when the informal channel is doing most of the actual work of the team and the formal channel is, in effect, a ratification mechanism for decisions already made informally. When this happens — and in my experience it happens often — the geographically distant team member is not really a team member. They are an output absorber. They receive the decisions of the central group and execute against them. They are not consulted on the conditions that produced the decisions, because the conditions are conveyed in conversations they were not in.

The Time-Zone Tax

The other variable that geography produces, and that the management literature treats too lightly, is time. A U.S.-India team faces a time-zone gap of nine and a half to twelve and a half hours. A U.S.-Poland team faces a gap of six to nine. A U.S.-Brazil team can sometimes coordinate around a single workday with effort. None of these gaps is intrinsically a problem. The problem is the question of who absorbs the cost of the gap, and the answer to that question is structural in a way the diversity-and-inclusion vocabulary has not yet developed a clean way to talk about.

The research literature on distributed teams — Catherine Cramton’s work on what she calls the mutual knowledge problem, the ACM literature on cross-time-zone collaboration, the broader body of work on virtual team performance — converges on a consistent finding. The convention of rotating meeting times across time zones, which is the recommended practice in the management literature, is rarely actually rotated in operating teams. What happens instead is that the team adjusts its standing meeting time to be reasonable for the highest-paid members of the team, and the lower-paid members of the team take the inconvenience. Indian developers working with U.S. teams routinely take meetings late into their evenings. Polish developers working with U.S. teams routinely take meetings into the late afternoon and early evening, after their counterparts have eaten lunch. The convention of fairness is asserted in the management documentation. The practice of unfairness is asserted in the calendar. The calendar is the document that describes what the team actually values, because the calendar is the document the team is actually living by.

The time-zone gap then compounds the proximity gap. The team member taking the meeting at the end of a long day is, by the time the meeting starts, less present, less able to contribute, and less likely to push back than the team member taking the meeting at ten in the morning. The asymmetry of participation is not a function of individual capability; it is a function of when, in a given person’s day, the meeting is being held. The outcome of the meeting, predictably, reflects the participants who were freshest. Those participants are, predictably, the ones whose time zone the team has implicitly privileged. The decisions that emerge from the meeting are then framed, in the meeting’s minutes, as team decisions. They are not team decisions. They are decisions arrived at by the segment of the team that was most awake and most able to talk. The remainder of the team is then asked to execute the decisions they did not have the bandwidth to shape.

The Operational Case for a Different Approach

I do not want to leave the description at the level of grievance, because the description, if it stops at grievance, is just one more iteration of a complaint the management literature has been hearing for two decades and has been comfortable not acting on. The reason the literature has been comfortable is that the existing arrangement is producing the numbers the existing arrangement is rewarded for producing. Until the numbers shift, the practice will not shift. The case for the practice shifting has to come from the demonstration that the numbers, when measured correctly, already favor the alternative.

The Gallup Q12 meta-analysis — an aggregation of seven hundred and thirty-six studies covering three and a half million employees in ninety countries — provides one version of the demonstration. Teams in the top quartile of engagement, as measured by the twelve-item instrument the analysis is based on, show twenty-three percent higher profitability, seventy-eight percent lower absenteeism, eighteen to forty-three percent lower turnover, and sixty-three percent fewer safety incidents than teams in the bottom quartile. The true-score correlation between engagement and a composite performance metric is approximately 0.49. For organizational research, a correlation of that magnitude is unusually strong. Gallup has estimated that declining global engagement cost roughly four hundred thirty-eight billion dollars in lost productivity over the most recently measured year. The numbers are not soft. The numbers are harder than the numbers most companies use for capital allocation. The fact that the numbers are not driving capital allocation toward engagement-supporting practices is a fact about corporate decision-making, not a fact about the numbers.

Amy Edmondson’s work on psychological safety, and Google’s subsequent Project Aristotle study of more than a hundred and eighty teams within its own organization, has produced a parallel finding. Psychological safety — the team-level confidence that one can speak up about errors, questions, or alternative approaches without being penalized — emerged as the single most important predictor of team effectiveness, ahead of individual talent, skills, or seniority. Edmondson’s research on hospital teams contains a finding that is initially counterintuitive and then, on reflection, obvious: higher-performing teams report more errors, not fewer, because safety enables the reporting that enables the correction. The number that distinguishes the high-performing team from the low-performing team is not the number of errors made; it is the number of errors named. The naming is the precondition of the correction. The correction is the precondition of the performance. The Gallup data shows that only three in ten employees strongly agree that their opinions count at their work. The capacity for organizational learning is, on this evidence, a rarer resource than any individual technical skill the same organizations claim to be selecting for.

The convergence point with the geography argument is direct. A globally distributed team in which the time-zone tax falls predictably on the lower-paid members, and in which informal decisions get made in proximity to leadership, is a team in which psychological safety, by structural design, is unevenly distributed. The team members who can speak up are the ones who are physically close and time-zone aligned. The team members who cannot are the ones who would, in a properly designed team, be the source of the error reporting that produces the learning. The team’s capacity for self-correction is, in effect, geographically truncated. The corrections happen, when they happen, on the side of the team that already has the advantages. The remainder of the team is structurally silenced not because its members lack the courage to speak but because the conditions for the speech to register are not in place. The team operates at a fraction of its possible learning rate. The fraction is the price of the geographic arrangement.

The Brittany Bond Finding

There is a more recent and more specific piece of research that ties the geography argument to the broader argument the series has been making. Brittany Bond, at Cornell’s ILR School, working with administrative data on performance ratings in a large company, found that employees who were nominated by their managers for top performance ratings but who were denied those ratings because of forced-distribution caps — the practice of requiring that a fixed share of the team be placed in lower ratings buckets regardless of underlying performance — were at least thirty-four percent more likely to leave the company voluntarily, even when receiving larger bonuses in compensation. A parallel simulation study published in the same period found that forced ranking misclassifies approximately a third of employees under ideal conditions, with the misclassification rate rising above half under realistic conditions, and that more than half of the variance in ratings is attributable to rater biases rather than to the underlying performance the ratings claim to describe.

The implication for geographically distributed teams is direct. Forced-ranking systems, when applied across a team in which some members enjoy proximity and time-zone advantages and others do not, are mechanically biased against the disadvantaged segment. The segment whose contributions the team is least able to surface, because of structural conditions the segment does not control, is the segment whose performance the rating system will rate lowest. The lowest-rated segment will then be the segment most exposed to elimination in the next round of restructuring. The restructuring, when it happens, will be described in the language of performance management. The performance management language will be operating on top of a measurement system whose biases are documented in the research literature the management of the company will not have read.

There is a further finding worth surfacing, because it ties the geography argument to the neurodiversity argument Essay 18 was concerned with. A 2023 study by Hartman and colleagues on autistic employees in technical organizations found that autistic employees were measurably more likely than their neurotypical counterparts to identify and report inefficient processes — precisely the kind of self-correction signal Edmondson’s work shows to be the precondition for team learning. The people best positioned to flag the problems are, in many measurement systems, the people most penalized by the rating systems being used to evaluate them. The geographically remote and the cognitively neurodivergent are not, mechanically, the same population. The structural position they occupy in the contemporary corporate measurement system is, mechanically, quite similar. The system selects against its own correction mechanism. The selection is automatic. The cost compounds.

What Working Examples Show

I want to spend a paragraph or two on the counter-examples, because the counter-examples exist and they matter. They are not numerous. They are not the dominant practice. They are present.

Microsoft, in the course of its reorientation under Satya Nadella, shifted its internal metric framework from engagement — a generic survey-based aggregate — to thriving, a measure defined as being energized and empowered to do meaningful work. The instrument now combines daily pulse checks, monthly live events, and behavioral telemetry like collaboration patterns, after-hours work, and focus time. The framework is designed to surface the root causes the older engagement instrument could not, and the structural conditions producing them, with the same analytical sophistication the company has long applied to its customers. Whether the new framework is fully realizing its potential is a separate question. The shift in measurement orientation is real, and the published descriptions of the program indicate a level of internal commitment to understanding employee experience that the prevailing engagement-survey norm does not require.

Patagonia is the smaller but clearer case. The company has, under its founder and current ownership structure, made a series of decisions that explicitly subordinate financial optimization to environmental and labor mission, including on-site childcare, paid environmental internships, flexible scheduling, and the declaration, in 2022, that the company’s only shareholder is the planet. Patagonia appears at the top of independent reputation rankings consistently. The companies that have maintained their diversity and inclusion commitments have, on the most recent reputation survey data, shown reputation score increases; the companies that have publicly rolled those commitments back have shown declines. The data points in the direction the operational research the rest of this essay has cited also points. The companies that treat their people, broadly construed, as a source of value rather than as a cost to minimize, perform better on the measures that capture long-term institutional health, and worse on the measures that capture only short-term cost optimization.

The Sam’s Club turnaround under John Furner, which Essay 20 cited in a different context, is worth returning to here because the geography of the turnaround is instructive. The redesign Furner led — reducing SKU complexity by twenty-five percent, simplifying work groups from twenty-seven to four, eliminating an overnight shift, cutting five thousand legacy tasks, raising wages by five to seven dollars per hour — was an operational redesign, not an HR program. The distinction is the distinction Zeynep Ton has emphasized in her work, and it is the distinction this essay has been circling. The improvements in productivity, turnover, and sales that followed the redesign were the result of changes to how work was structured, not the result of better training or wellness programs. The internal resistance to the wage increase came, instructively, from the finance and HR functions whose measurement systems could see the cost of the increase but could not see the avoided cost of not making it. The measurement systems that resisted the change were the same systems whose blindness Ton’s work was diagnosing. The redesign succeeded because the leadership was willing to act against the measurement system’s recommendation. The leadership was rewarded with a promotion to run the larger Walmart organization. The lesson, for those reading the case, is that the right operational answer is sometimes available and is sometimes implemented. The case is rare enough to be a case. The rarity is the diagnosis.

The Pittsburgh Paradox

I want to close, before the final section, with a local observation, because the geography of dignity does not stop at the international border. It also operates within cities. The city I live in is an example of the broader pattern at a smaller scale.

Pittsburgh, by certain measures, is in the middle of a substantial economic boom. The tech ecosystem — anchored by Carnegie Mellon, by the University of Pittsburgh and the UPMC health system, by the autonomous-vehicle and AI-research firms that have located in the region to be close to the universities — has attracted upward of ten billion dollars in cumulative investment. The boom has produced new neighborhoods, new restaurants, new transit lines, and new wealth, much of it concentrated in the corridors that run from Oakland through East Liberty into the Strip District. By those measures the city has, as the local marketing language has it, reinvented itself.

By other measures, the same city has child poverty rates of twenty-five percent or higher in its majority-Black neighborhoods, which sit a few miles from the tech corridors and across institutional networks that do not, by and large, connect to them. The two economies coexist in the same city. They are not the same economy at different stages of development. They are different economies, with different rates of growth, different access to capital, different connections to credentialing institutions, and different prospects for continued investment. The labor arbitrage Essay 21 has been describing at the international level is happening at the local level as well, with the difference that the participants are not separated by passports but by a few miles of road and several decades of structural disinvestment. The geographically remote team member on the other side of the ocean and the geographically adjacent worker on the other side of the river are facing, structurally, a more similar situation than the language of global teams usually concedes. The corporate apparatus that decides which work goes where, and at what price, operates on the same calculation in both cases. The calculation is opaque in both cases. The opacity is the apparatus.

Closing

The question this essay opens with — in a global company, who stays human? — admits of only two answers. The first is that everyone does. The second is that no one does. There is no in-between. The arrangement in which some team members are treated as humans whose dignity matters and others are treated as labor inputs whose dignity is not the company’s problem is not a stable arrangement. It is the unstable arrangement that produces the conditions the rest of this series has been describing. The home-country employee who learns to participate in the dehumanization of the offshore colleague is the home-country employee who will, in time, find their own dehumanization arriving on the schedule the home-country measurement system had been forecasting all along. The reposting of my role in Wrocław is one small version of that arrival. The next version is somebody else’s.

The case for dignity across geography is not a moral case in the sense the corporate communications department uses the word. It is a structural case. The teams that distribute dignity unevenly do not, by the available evidence, perform better than the teams that distribute it more evenly. They perform worse on every long-horizon measure the research literature has been able to capture, and they perform better only on the short-horizon cost measure the financial-statement format makes most visible. The cost measure is the measure the bad-jobs equilibrium has been organized around for two generations. The other measures are the ones that catch up, eventually, in the form of attrition, error rates, customer dissatisfaction, regulatory exposure, and the slow erosion of the institutional knowledge that took decades to build and that no quarterly report contains a line item for. The catch-up arrives later than the cost. The catch-up arrives. The institutions that were not paying attention to anything but the cost are the institutions least equipped to recognize the catch-up when it does.

The geography of dignity is not, finally, a question about other countries. It is a question about whether the organization can extend the recognition of human standing across distances it is not directly experiencing. The organizations that can do this consistently are the organizations that have survived multiple business cycles, weathered downturns without destroying themselves, and produced the long-horizon returns the prevailing management discourse claims to care about. The organizations that cannot are the organizations whose names appear on the layoff trackers, in the financial-distress filings, in the regulatory enforcement actions, and in the human stories that begin, generally, with the words my role was eliminated. The geography is not separate from the rest of the analysis. The geography is the analysis at scale. I want to be careful about what conclusion this points to, because the tempting conclusion is the absolute one — everyone stays human or no one does — and the absolute conclusion is not one I hold. The technology is not going anywhere, and it should not. The tools I have spent a career building are, deployed with any care at all, exactly the kind of leverage that lets fewer people do less drudgery for more value. The question was never whether the machines come. The question is what happens to the people standing where the machines are aimed while the transition is underway. History is not reassuring on this point. The Industrial Revolution is the standard consolation — people found new jobs eventually — and the consolation only works if you skip the part Harari’s Nexus lingers on: the generations ground up inside the transition itself, the decades of immiseration and social collapse that were not a natural price of progress but the product of decisions made by people in power who gave no serious consideration to the people underneath them. Eventually is doing a great deal of work in that consolation, and the people it erases are the people who lived and died inside the eventually. The obligation, then, is not to stop the transition. The obligation is to run it as if people are standing in the blast radius, because they are. That is a lower bar than everyone staying human. It is also, on the present evidence, a bar the corporate apparatus is not clearing — and the distance between the bar and the apparatus is the distance this essay has been measuring.

← Essay 20: Transparency as Self-Defense Essay 22: Reclaiming the Narrative → View Full Archive