The Last Human Job
The Work of Connecting in a Disconnected World
by Allison Pugh
2024
The Last Human Job is social theory that's deeply grounded in interviews and ethnographic research that author Allison Pugh conducted with people in a wide variety of caregiving professions, including hair dressers, home health aides, teachers, chaplains, and medical doctors. Pugh has two goals - first to name and describe the work of making an authentic connection with another person, and second to show how this necessary labor (and the workers who do it) are being degraded by various attempts to accelerate and automate it.
Pugh identifies what she calls connective labor, which is work we do while interacting with another person. It's the work of listening to them and responding back to them so that they feel understood, they feel seen. Care work often requires connective labor as part of the job - but so do lots of other kinds of work too. Pugh thinks we often fail to recognize this as a kind of work, and she thinks that for many jobs, how well the worker can connect with the client determines how well the actual job will go. A good hairdresser, for example, isn't just good at styling hair, but at knowing how each person wants their hair to be styled, understanding how they wants others to see them. For some jobs, the 'real' work can't happen at all unless the worker can make a connection first.
For one person to see another can sometimes be a powerful, profound experience, for both of them. It can also be draining for the worker. Jobs that require connective labor often have high rates of burnout. People probably vary in their innate talent for making connections, but Pugh thinks it's also a skill that can be learned, practiced, and improved on. Ultimately Pugh thinks that it's the organization of the workplace, much more so than the qualities of the individual worker, that determine whether they are able to successfully connect with their clients, and whether they can do so consistently and sustainably, without using themselves up.
Pugh identifies three characteristics of workplaces that make it possible for workers to successfully provide connective labor to clients. First, they're set up so that workers are supported by leaders and supervisors who believe in the value of connection, mentors who can help them learn what to do, and peers who can act as sounding boards. To keep doing this kind of work, people need advice, encouragement, the ability to admit mistakes and learn from them. Second, workplaces support connective labor by making relationships an official priority. They recognize that they need to make a connection first, before they can do their 'real' work.Third, they support this work with resources - enough workers to meet client demand and enough time with each client to do the job well.
That last requirement gets at the heart of what Pugh thinks is the biggest threat to our collective ability to connect with one another. Connective labor often costs too much for the people who need it most to afford it, it pays too little for many of the workers to keep at it. There is too much work to be done, too many people in need, and not enough staff or time to do it all. And meanwhile workers are under constant pressure to see more people, spend less time with them, do the job faster, get more done, to do it without actually connecting at all. Collectively, these pressures threaten to reserve high-quality connective labor for the rich, while everyone else gets worsening service. And the worse the service becomes, the more tempting it is to replace it with some form of automation.
Pugh finds a few really common arrangements for workplaces that deliver connective labor to clients. There are mission-driven facilities located in poor areas, where staff are called on to be heroes, and there's always a backlog of unmet need. Workers here are often able to connect well, but become burned out from simply having to do too much, for too long each day. Corporate facilities have more resources and allow workers to have a home-life separate from their work, but they schedule lots of very short appointments. Workers are still overworked, and they're denied the ability to make meaningful connections. The personal service model is a luxury for the well-off. Workers have time and resources, but get treated like servants, and often feel bad about the people who can't afford their help. Set-ups that provide concierge-level resources to workplaces that serve the neediest people do exist, but they're much rarer.
From the outside, it seems like there wouldn't be quite so much unmet need if we didn't organize this work in such a way that it continuously burns people out and uses them up. I have to suspect that people vary somewhat in their need for connective labor. We might not be able to give everyone the concierge treatment, but not everyone needs it. But also, the people who need it most are probably the least able to afford it. Again from the outside, government funding seems like the obvious solution to this mismatch. Especially because programs like housing-first approaches to homelessness, or assigning care coordinators to patients with complex conditions, or experiments with unconditional basic income always produce better results and cost less long-term than the way we usually do things.
The threat to connective labor comes primarily from corporations, governments, and other leaders who all want to find ways to spend less money on it. Inevitably, these plans involve some form of automation. By automation, Pugh does not primarily mean things like replacing therapists with genAI chatbots - although she does explicitly consider that companies might try to deprive poorer clients of any human contact at all. Automation here refers to any attempt to use technology or organization to divide work into smaller components so that they can be performed more efficiently. Pugh argues that leaders are currently trying to do to care work and connective labor what the Industrial Revolution and the assembly line did for manufacturing.
Automation promises to save workers from burnout by helping them work more efficiently, but mostly worsens it by raising expectations for how much they should be able to do while draining their work of autonomy and meaning - the things that make it satisfying. And despite their promises, automation schemes often actually add tasks and take more time - but that time is moved away from clients and onto the work of following and documenting the automation protocols.
The most common forms of automation for this kind of work are scripting and counting. By scripting, Pugh means a whole variety of ways that work can be standardized by employing scripts, templates, manuals, checklists, flowcharts, etc. The goal is to make it go faster by making it go the same way every time (this is also a hedge against incomplete or low-quality interactions). But of course clients' lives and problems aren't scripted, and might need more time or a different approach than has been allotted to them.
Counting refers to various forms of data collection. It's meant to track what's being done (especially useful for billing purposes) and to demonstrate that it's being done well. Customer satisfaction surveys and standardized tests for students are two common examples. Counting imposes its own problems - it's time consuming, it can focus attention away from things that are important but difficult to quantify, and it tempts workers and bosses alike to focus on getting the right number rather than doing whatever the number is supposed to represent.
The social theories we use as the basis of sociology were all about changes in the way people interact, know each other, and form communities in the wake of mass urbanization as people moved into cities, and about the ways they were affected by industrialization and the reorganization of work. Pugh is very much addressing those same topics, as they are relevant in the contemporary context.
Pugh identifies what she calls connective labor, which is work we do while interacting with another person. It's the work of listening to them and responding back to them so that they feel understood, they feel seen. Care work often requires connective labor as part of the job - but so do lots of other kinds of work too. Pugh thinks we often fail to recognize this as a kind of work, and she thinks that for many jobs, how well the worker can connect with the client determines how well the actual job will go. A good hairdresser, for example, isn't just good at styling hair, but at knowing how each person wants their hair to be styled, understanding how they wants others to see them. For some jobs, the 'real' work can't happen at all unless the worker can make a connection first.
For one person to see another can sometimes be a powerful, profound experience, for both of them. It can also be draining for the worker. Jobs that require connective labor often have high rates of burnout. People probably vary in their innate talent for making connections, but Pugh thinks it's also a skill that can be learned, practiced, and improved on. Ultimately Pugh thinks that it's the organization of the workplace, much more so than the qualities of the individual worker, that determine whether they are able to successfully connect with their clients, and whether they can do so consistently and sustainably, without using themselves up.
Pugh identifies three characteristics of workplaces that make it possible for workers to successfully provide connective labor to clients. First, they're set up so that workers are supported by leaders and supervisors who believe in the value of connection, mentors who can help them learn what to do, and peers who can act as sounding boards. To keep doing this kind of work, people need advice, encouragement, the ability to admit mistakes and learn from them. Second, workplaces support connective labor by making relationships an official priority. They recognize that they need to make a connection first, before they can do their 'real' work.Third, they support this work with resources - enough workers to meet client demand and enough time with each client to do the job well.
That last requirement gets at the heart of what Pugh thinks is the biggest threat to our collective ability to connect with one another. Connective labor often costs too much for the people who need it most to afford it, it pays too little for many of the workers to keep at it. There is too much work to be done, too many people in need, and not enough staff or time to do it all. And meanwhile workers are under constant pressure to see more people, spend less time with them, do the job faster, get more done, to do it without actually connecting at all. Collectively, these pressures threaten to reserve high-quality connective labor for the rich, while everyone else gets worsening service. And the worse the service becomes, the more tempting it is to replace it with some form of automation.
Pugh finds a few really common arrangements for workplaces that deliver connective labor to clients. There are mission-driven facilities located in poor areas, where staff are called on to be heroes, and there's always a backlog of unmet need. Workers here are often able to connect well, but become burned out from simply having to do too much, for too long each day. Corporate facilities have more resources and allow workers to have a home-life separate from their work, but they schedule lots of very short appointments. Workers are still overworked, and they're denied the ability to make meaningful connections. The personal service model is a luxury for the well-off. Workers have time and resources, but get treated like servants, and often feel bad about the people who can't afford their help. Set-ups that provide concierge-level resources to workplaces that serve the neediest people do exist, but they're much rarer.
From the outside, it seems like there wouldn't be quite so much unmet need if we didn't organize this work in such a way that it continuously burns people out and uses them up. I have to suspect that people vary somewhat in their need for connective labor. We might not be able to give everyone the concierge treatment, but not everyone needs it. But also, the people who need it most are probably the least able to afford it. Again from the outside, government funding seems like the obvious solution to this mismatch. Especially because programs like housing-first approaches to homelessness, or assigning care coordinators to patients with complex conditions, or experiments with unconditional basic income always produce better results and cost less long-term than the way we usually do things.
The threat to connective labor comes primarily from corporations, governments, and other leaders who all want to find ways to spend less money on it. Inevitably, these plans involve some form of automation. By automation, Pugh does not primarily mean things like replacing therapists with genAI chatbots - although she does explicitly consider that companies might try to deprive poorer clients of any human contact at all. Automation here refers to any attempt to use technology or organization to divide work into smaller components so that they can be performed more efficiently. Pugh argues that leaders are currently trying to do to care work and connective labor what the Industrial Revolution and the assembly line did for manufacturing.
Automation promises to save workers from burnout by helping them work more efficiently, but mostly worsens it by raising expectations for how much they should be able to do while draining their work of autonomy and meaning - the things that make it satisfying. And despite their promises, automation schemes often actually add tasks and take more time - but that time is moved away from clients and onto the work of following and documenting the automation protocols.
The most common forms of automation for this kind of work are scripting and counting. By scripting, Pugh means a whole variety of ways that work can be standardized by employing scripts, templates, manuals, checklists, flowcharts, etc. The goal is to make it go faster by making it go the same way every time (this is also a hedge against incomplete or low-quality interactions). But of course clients' lives and problems aren't scripted, and might need more time or a different approach than has been allotted to them.
Counting refers to various forms of data collection. It's meant to track what's being done (especially useful for billing purposes) and to demonstrate that it's being done well. Customer satisfaction surveys and standardized tests for students are two common examples. Counting imposes its own problems - it's time consuming, it can focus attention away from things that are important but difficult to quantify, and it tempts workers and bosses alike to focus on getting the right number rather than doing whatever the number is supposed to represent.
The social theories we use as the basis of sociology were all about changes in the way people interact, know each other, and form communities in the wake of mass urbanization as people moved into cities, and about the ways they were affected by industrialization and the reorganization of work. Pugh is very much addressing those same topics, as they are relevant in the contemporary context.
"Hair dressers, home health aides, teachers, chaplains, and medical doctors" sounds like a field that converges on "homemaker," historically an undervalued field of economic human endeavor but one that naturally resists automation for exactly that reason. Not sure Pugh even gestures toward that third rail but it makes me think about whether the disappearance of most "jobs" (manufacturing, media, bureaucratic "email jobs") means the only labor left to humans will look a lot like parenting.
ReplyDeleteIt's important to note that the deprecating on this sort of personal touch doesn't just come in the name of corporate profits. In healthcare, providing care for the most people in the name of access often means shorter visit times.
ReplyDeleteYeah, there are human capacity constraints, only so many metaphorical "hugs" a pair of arms can administer in a given time period. I think the follow through on my initial comment is on the scripting / counting / billing automation loop: anything that can be easily quantified and scripted already is, the last things that will go unscripted are either very difficult to script, undervalued economically or both.
ReplyDeleteWhich is to say, looks like "mother."