Unionized Software Engineers - Is the Social Cost spread out disproportionately?

Today, there was an article announcing a new workers union for all employees working at Alphabet, Inc., Google’s parent company, the Alphabet Workers Union. The union is constituted by about 400 employees, an extremely small percentage of the total workforce of nearly 250,000 people. The union dues are 1% of total compensation.

We still have discrimination, harassment, & retaliation at Google. Over half our workforce are TVCs—paid less, w/fewer benefits, for often the exact same work.

We deserve meaningful control over the projects we work on & the direction of this company.

Here’s where it gets interesting: The union is a minority union. As the union doesn’t have a majority of all the employees as members, there is no basis for them to walk up to Alphabet’s management and demand to negotiate a Union contract. Instead, they are going to use their union as a more abstract entity that will be used for other purposes. The question that came up when I was discussing this with a few co-workers was whether the social cost of this particular type of union is spread out disproportionately between the employer and the union members?

I have been reading up about unions after reading The Box, the book about the introduction of the shipping container and how it took over the world. The most fascinating part of the book for me, was the explanation of the “transition” period: the period when shipping companies, ports, and shipping users were deciding whether to move from break-bulk shipping to container shipping. Break-bulk shipping involves the packing and unpacking of material at each intermediate port. These packing, unpacking, re-packing steps were carried out by dockworkers. These dockworkers were a part of unions across the world. In some countries, they embraced the shipping container, deciding to improve their skills and be a part of the crew that would help initiate the transition. In other countries, dockworkers unions where opposed to the idea and did absolutely everything they could to ensure that the shipping container would not be introduced at the port that they were working at.

By “absolutely everything”, I mean a set of general strikes, restrictive union agreements, and increases in wages. One of the most fascinating clauses that one union agreement had was a clause that each hatch on a ship would have to be manned by a minimum number of people, irrespective of the actual amount of cargo that will be moved through that hatch. This lead to cases were the minimum was 16 people, but only 8 people would be working at any given time. To me, this indicated a broken system that was not ready to adopt new technology. Was my assessment fair? No, it was wrong.

Organizing to collectively bargain is a major part of the reason that a lot of workers have more leverage over their employers actions. Unions also have power in political elections, which gives them more attention from policy makers, and enables them to push for their policy preferences. This system has it’s advantages and can not be dismissed out-of-hand. This post will explore what is really going on here, look at the issue from a few perspectives, and try to untangle the various threads.

Disproportionate social cost

From society’s perspective, the negotiation between a union and the employer is a near-zero-sum game: Union membership has some major costs, mainly, union dues that one must pay constantly, the possibility of strikes when the union as a whole or the union leadership is unhappy with the work conditions, and the possibility of loss-of-pay during the strike while union members are paid out a lower stipend or given rations, using the fund maintained using their union dues.

When the union is happy, the cost is almost zero for both employee and employer. There is the union dues cost, which is balanced out by the time that the employer spends to ensure that their policies are not exploitative and will not anger the union enough to cause a strike.

When the union is unhappy, for a traditional union, the cost is still evenly spread out: The company loses their revenue, their ability to fill orders and a major drop in production, which will affect their yearly output targets. Union members bear an equal cost, as they are essentially out of a job and must now start managing with a stipend or rations; in some cases, they also need to physically be at the picket line and spend their time there picketing. Over-all, I would say that the cost of a traditional union during both conditions is proportionately spread out.

Selby Jr’s Last Exit To Brooklyn is a novel that captures the mentality of a particularly lethargic Union member (Harry) from several decades ago.

Now, take the case of the Alphabet Union: They don’t have any contract. But they do reserve the right to not work on projects that they don’t agree with:

All aspects of our work should be transparent, including the freedom to decline to work on projects that don’t align with our values. We need to know the impact of our work, whether it’s on Alphabet workers, our communities, or the world.

Values, Alphabet Workers Union

At this point, the costs are no longer proportionately spread out anymore. The employer can’t do much in this case (certainly not an employer like Google) because these workers are not going on strike as a majority union. They are simply refusing to work on a particular project. Let’s call these projects that some people don’t want to work on, “morally disagreeable projects”.

For a large-scale company like Google, it will be extremely easy to move people around and find people to work on any project, their pay-scale and the prestige of working at that employer is high enough to balance out the “moral dilemma” caused by these projects. On the other hand, a smaller-scale company which wants to get involved in a morally disagreeable project will find it very hard to staff that project.

In general, an employer who wants to work on such a project has a few options to move forward:

  1. Staff the project using people who don’t mind working on it OR
  2. Stop working on the project

Option (1) will always be feasible for large-scale companies; they will certainly work for companies like Alphabet. Option (2) will be the most obvious option for smaller scale companies. If the company depends on this project for it’s survival, it might go under. This in-turn will have a ripple effect, that leads to a “bad rep” (a bad reputation) for places where such projects can not be undertaken. Leading to newer companies attempting to work on these morally disagreeable projects either not coming up at all or coming up in a different location and breaking up the Silicon Valley hegemony on start-ups.

Either way, it is quite clear to me that the existing smaller-scale companies and future start-ups, in the location where organizing similar to the Alphabet Union becomes popular will bear the brunt of the burden. This disproportionate burden on one party leads me to believe that there is a structural problem with the idea of a union where the costs are disproportionately shared.

Structural-concerns

As we have seen, the Alphabet Union is not going to negotiate a Union contract. Instead, they are going to exist as a body that is going to work on the sidelines. Google’s current rhetoric seems to be a complete lack of recognition for the union: Employers can insist that they will only deal with employees on an individual basis, as hinted at in this New York Times article:

In response, Kara Silverstein, Google’s director of people operations, said: “We’ve always worked hard to create a supportive and rewarding workplace for our work force. Of course, our employees have protected labor rights that we support. But as we’ve always done, we’ll continue engaging directly with all our employees.”

Although they will not be able to negotiate a contract, the Alphabet Workers Union can use other tactics to pressure Google into changing its policies, labor experts said. Minority unions often turn to public pressure campaigns and lobby legislative or regulatory bodies to influence employers.

NYT article announcing the union

In the lack of a union contract, the benefits for membership are very little. One gains some power from the fact that this block of employees might work together to push the employer into some difficult arguments in the public sphere, but this would be no different from the public pressure campaigns that employees at several software companies have been running over the past year as facial recognition and software for border control agents have started becoming hot-button issues with this group of people.

Furthermore, one of the biggest problems that experienced organizers identify and talk about is the lack of incentive for the highly-skilled, focused-on-work employees to join traditional unions. These employees tend to have good relationships with their manager, good wages and a generally good opinion of their employer. They are the hardest to get to join the union, and they are the most important for the union, because in their absence, the union does not have legitimacy in the employer’s perspective. McAlevey talks about this at length in her interview on the Ezra Klein Show.

I believe that most software engineers work in firms where the difference in skill varies only by their grade, and rarely has huge variations within peers who are on the same grade. In this situation, either all the people from a given grade will have an incentive to join the union OR they will not join the union. My bet is that the membership numbers for the Alphabet Union will stall and they will not be able to sign-up even 10% of the total Alphabet employee base.

Finally, the other major structural concern that I came up with was the difference in the union’s use-case. Traditionally, unions have been used by workers who have been exploited by their employer. The ability to bargain collectively gives these workers the kind of leverage that they can’t get if they were to negotiate their work hours or pay with their direct manager in a 1-1. This technique has been effectively used by factory-floor workers, dockworkers, teachers to get better pay and better work hours. In all these cases, the underlying issue is related to a concrete Labor Law issue; this concrete connection makes the reasoning objective. On the other hand, a morally disagreeable project, is ill-defined and completely subjective. The lack of objectivity makes it very hard for the employer to predict which employees will not work on what projects (HR departments frown on discussions about political leanings at work, although there is wide variation on this between companies).

What is a “morally disagreeable project”?

(The subjectivity of this definition might be abundantly clear to you. If so, you can skip this section.)

The moral nature of a project and whether it agreeable to one or not depends on their political leanings and personal philosophy. This evaluation is bound to vary from one person to the next, and there is no real “control” group here that can be used as a litmus test to decide whether something is disagreeable to a majority or not (Elections are the closest we come to implementing this mechanism, and we often don’t get the result that we want, or even the result that the majority wants if the system has structural problems).

Going further, I argue that one would be able to find an argument against any company that harms either society or the environment. Several existing arguments can be re-framed as arguments that harm people or the environment.

In the case of Amazon, the impact of 2-day shipping on the environment has been widely studied. This Vox video condenses a bunch of the research and concepts into a short snippet.

It will get tougher for employers to adjudicate whether a given project will be morally disagreeable to the team that will work on it. The best case result of this is an employer whose limits match exactly with their employees. The worst case is one where the employer and the employees are on completely different sides of the spectrum, leading to either pandering to ensure that attrition stays low or doing what the employer wants to do at the risk of losing talent, due to a belief that what the employer wants to do is worthwhile.

Facebook is already facing challenges like this one and as Mark Zuckerberg puts it, “We should remember that Facebook’s employees are much further to the left compared to Facebook’s users”. As the employer, Zuckerberg and Facebook, the company need to focus on their users. But they are also caught in a bind with their own employees unwilling to work on a platform that they believe is empowering the people that they don’t want to empower. Casey Newton and Ezra Klein flesh this out further on the podcast, The Interface.

The border line between agreeable and disagreeable is shifting frighteningly fast. The discourse has moved a long way away from how we were talking about climate change or fossil fuel consumption reduction or even fossil fuel production compared to 2010, only 10 years ago. The change over the next 10 years might not be as dramatic, but there is no way to tell. Organizations that like to have 10 or 20 year plans and “roadmap” slide decks are going to have a nightmarish time figuring out how much things will change.


Over the coming months, we will see the saga of the unionized software engineers and other Alphabet employees play out. We will also see what happens when the moral alignments of an employer and their employees does not line up perfectly, by looking at the crisis that seems to be brewing inside Facebook. Both of these are new phenomena that are caused due to a simultaneous increase in the irreplaceable nature of some employees in these companies and the increase in the scale of these companies and the incredible variety of people that their products touch.