The only thing less surprising than a union losing a big election for representation right in the Deep South is that the loss generated a firestorm of takes. Smart thinkers on the Left like Nelson Lichtenstein, Jane McAlevey, John Logan, and Rich Yeselson have already weighed in with their thoughts, and many more discussions are taking place online and offline regarding what went wrong in the Retail, Wholesale and Department Store Union (RWDSU)’s loss last week in the National Labor Relations Board (NLRB)–conducted election at Amazon’s facility in Bessemer, Alabama. The mood is understandably somber in assessing the fallout, but it’s also surprisingly bitter for a campaign that was a long shot from the moment it was unexpectedly revealed to the world late last November.

All parties acknowledge that the RWDSU faced both enormous structural disadvantages inherent in any organizing drive under the National Labor Relations Act, and the unique challenges of attempting to unionize the country’s most powerful open shop, which wields unlimited resources and unmatched surveillance capabilities. But many of the postmortems of the election dive deep into criticisms of the union’s organizing strategy.

McAlevey, for example, takes the union to task on multiple fronts, including for supposedly being surprised at Amazon’s successful quadrupling of the bargaining unit in NLRB proceedings, for declining to conduct house visits of eligible voters, and for failing to prioritize outward displays of union support on the shop floor. Yeselson adds that the union foolishly proceeded to the election without enjoying supermajority support in the inflated unit, leaving organizers scrambling to build momentum when campaigns should be busy inoculating those workers against the inevitable anti-union campaign.

These criticisms may all have merit. They are also irrelevant to the broader issue facing the labor movement.

First, this is a process undertaken through the election machinery of the NLRB. While agonizingly slow, a major component of that process is the built-in challenge to the election results. Any unfair labor practices or other sufficient forms of interference that Amazon committed during the election would likely set the results aside and trigger a second election. Unions that are granted “re-run” elections do not have great track records at later emerging victorious and, more importantly, finalizing a collective bargaining agreement with the offending company. Among many other details that are likely forthcoming, Amazon’s unprecedented defiance of the NLRB’s order to not erect its own private ballot box certainly suggests that the current result is unlikely to survive the union’s objections.

Second, unions lose elections constantly. They lose them when they do house visits. They lose them when workers wear buttons on the shop floor. And they lose them even when possessing a supermajority card count before filing a petition. Far more important than the elections that are lost are the elections that never see the light of day.

Union win rates in NLRB elections are actually much higher than they were in the middle of the century, but in those decades the labor board was conducting upwards of eight thousand elections per year (with as many as half a million workers annually casting ballots). 1967 alone saw over 340,000 workers organized through the NLRB process. In comparison, only 1,059 union-called elections were held in the labor board’s most recent non-COVID year, covering just shy of sixty-five thousand workers. But hey, unions won 70 percent of those elections in 2019, an all-time high since World War II.

The harsh truth is that the slow and cautious supermajority standard for proceeding to an election has only helped accelerate the process of union decline in the private sector. Even the billions of dollars spent on organizing in the John Sweeney era didn’t lead to any detectable uptick in elections, and the non-NLRB routes of organizing didn’t put a dent in union density. But there’s a common variable coursing through these trends, one that the supermajority standard evolved in reaction to: the hostility of American labor law to union organizing.

Experts have argued for decades that the NLRA system of union certification is broken, emphasizing the lack of restraints on employers from breaking the law and the structural advantages employers enjoy through the protracted secret ballot process. There is hard data that confirms these assertions. The two most important weapons that Amazon deployed against the union were perfectly legal under current law: 1) the near-daily use of captive audience meetings, which were declared lawful by the NLRB after the Taft-Hartley Act’s passage, and 2) the gerrymandering of the bargaining unit in the preelection hearings, which is made possible by employers’ legal right to standing in representation proceedings (an original sin of the Wagner Act) and was recently exacerbated by a Trump board decision.

The Protecting the Right to Organize Act, currently pending in the Senate, would make these tactics unlawful by declaring captive audience meetings an unfair labor practice and by removing employers’ ability to litigate the appropriateness of their workers’ bargaining units in the pivotal early stages. The PRO Act would make many more revisions to the NLRA, but these obstacles in particular (captive audience meetings and standing in representation proceedings) make organizing an Amazon warehouse virtually impossible, especially given the rapid turnover of the company’s workforce. Amazon’s anti-union tactics had a far greater impact on the final vote than any single strategy the union pursued or neglected.

While the fine details of labor law are often rightfully criticized as arcane or overly complicated, the truth is that workers understand the law just fine. They understand that when their employer can force them to sit for hours in a room and feed them an anti-union message while simultaneously confining the union’s pitch to a voluntary nature outside of the facility, the former party has power while the latter party does not. That imbalance, buttressed at every turn by current law, infects any interaction that a union will ever have with a potential convert. Nothing in the last seventy-plus years of history suggests we can simply organize our way out of this predicament; employers must be legally divested of these advantages just as they are in the vast majority of other industrialized nations.

Passing the PRO Act won’t be easy. With five Democratic senators currently uncommitted to its provisions, the Senate filibuster still intact, and business interests still yet to launch a major counteroffensive, passage may not even be likely. But the loss in Bessemer shows why labor law reform is so crucial and why it’s worth expending the energy needed to give it a chance of passing. The labor movement may not survive without it.