The Deal Is Not Yet the Contract

On September 24, 2023, the Writers Guild of America announced a tentative agreement with the Alliance of Motion Picture and Television Producers on a new Minimum Basic Agreement. Its negotiating committee called the deal “exceptional.” It also told members that final contract language still needed to be drafted, approval procedures remained ahead, and nobody was authorized to return to work.

Relief had arrived. Permission had not. For an industry built around satisfying endings, this was an admirably inconvenient distinction.

This analysis stays at that September 24 boundary. The announcement establishes an agreement in principle, not the contents of a completed contract, a ratification result or a record of enforcement. Those are separate things, and collapsing them would obscure what the writers were still being asked to do: wait for the language, examine the terms and make a decision.

What had actually been agreed

The committee’s September 24 statement defined the tentative agreement precisely: “an agreement in principle on all deal points, subject to drafting final contract language.” It credited members’ willingness to endure the strike, together with support from other unions, for producing the leverage that brought the companies to a deal. That is the negotiating committee’s account of what moved the negotiations, not an independently measured allocation of credit.

The next steps were explicit. Once the Memorandum of Agreement was complete, the negotiating committee would vote on recommending it to the WGA West board and WGA East council. Those bodies would then decide whether to authorize a membership ratification vote. They could also vote to end the strike at a specified time pending ratification, without removing members’ final say over contract approval.

The leadership votes were tentatively scheduled for Tuesday, if the language was settled. A comprehensive summary and the memorandum would follow those votes, along with meetings where members could assess the deal.

That sequence matters because representation does not erase the represented. A committee can negotiate an agreement and recommend it enthusiastically without possessing the membership’s decision in advance. The announcement describes a route from negotiation to judgment, not a request to treat applause as consent.

It also separates two actions easily confused in a headline. WGA picketing was being suspended immediately; the strike was not. Members were encouraged, if able, to join SAG-AFTRA picket lines. The instruction was not “everything is over.” It was that one form of activity had changed while the prohibition on returning to work remained.

A profession is not a succession of lucky breaks

To understand why the distinction between a deal and its terms deserves attention, look at what the union said was failing before the strike.

In its March 14 compensation bulletin, the WGA reported that 49 percent of TV series writers worked at contractual minimums in the 2021–22 season, compared with 33 percent in 2013–14. It also reported that median weekly writer-producer pay had declined 4 percent over the preceding decade, or 23 percent after adjusting for inflation.

These are the union’s published figures, offered as part of its bargaining case. They should not be mistaken for independently audited findings or stretched into a claim about every writer. Their relevance is more specific: the organization representing writers was describing deterioration that could coexist with employment and professional advancement.

The bulletin also identified shorter employment periods on streaming series and the separation of writing from production as problems. Its tables showed median employment of 20 weeks for streaming staff writers in 2021–22, against 29 weeks on network and CW series. A weekly rate cannot explain annual security when the number of paid weeks differs.

This is where the familiar cultural story about creative success becomes inadequate. Getting hired is an event. Making a living requires a pattern of paid work that continues beyond the event. A person can acquire a desirable credit without acquiring a dependable year.

That does not make a credit worthless, or mean every short production should last longer. It means the prestige of participation cannot answer questions about duration, compensation and what happens between jobs. An industry can produce work audiences love while leaving the people making it with a much less attractive calendar.

The May 1 strike announcement made the union’s diagnosis unusually sharp: the companies had created “a gig economy inside a union workforce.” The statement announced a strike beginning May 2 and connected the dispute to weekly employment, free work for screenwriters and artificial intelligence, among other issues. This was a bargaining argument about how the profession was being organized, not simply a request that the public appreciate writers more.

The exception can consume the promise

An earlier negotiating update shows why the committee’s insistence on finishing the language is more than ceremonial fussiness.

In its August 24 account of the negotiations, the WGA acknowledged movement from the companies while arguing that their proposals remained insufficient. It described proposed minimum staffing and duration guarantees in television as weakened by loopholes, limitations and omissions. It said a proposed second writing step for screenwriters covered only a narrow category, excluding writers beyond the first writers of original screenplays.

Those descriptions are the union’s characterization of an earlier company offer. They are not a neutral adjudication of that offer, and they are emphatically not the terms announced on September 24. The point is the kind of disagreement they reveal: both parties can discuss a protection while disputing who actually receives it.

A guarantee’s name does not determine its reach. Eligibility, exclusions and the period covered can change its practical value. “There is a minimum” leaves unanswered the question “a minimum for whom?” A protection may be real within its boundary and inadequate because of the boundary.

The August statement supplied a smaller, especially legible example. According to the committee, the companies proposed pension and health contributions treating members of two-writer teams individually, but not teams of three or more. Without evaluating the merits of that proposal independently, the distinction illustrates why contract assessment cannot stop at the phrase “individual contributions.” The number of people in a team changes who benefits.

None of this establishes that the September agreement retained those exclusions. It explains why the absence of final language on September 24 limits what can responsibly be celebrated as a specific achievement. The announcement supports confidence that the negotiators believe they have a strong deal. It does not yet let readers inspect the coverage of each protection themselves.

Data and AI are questions about control

The August update also described a dispute over streaming information. In the union’s account, the companies had offered to let six WGA staff study limited viewership data for three years, without allowing writers to be told how their projects performed or receive a viewership-based residual under that offer.

Again, this is an earlier proposal as described by one side, not a description of the eventual settlement. But it separates three things that the pleasant word “transparency” can conceal: someone gets access to information; the affected person learns something usable; compensation changes as a result.

Those are not interchangeable. Data can support bargaining without immediately changing a writer’s pay. Restricted access may have value without meeting the need that prompted the request. The question is not whether information exists somewhere. It is who may use it, for what purpose and with what consequence.

Artificial intelligence presents a related temptation to confuse a category with an answer. The August statement reported real discussions and movement on AI protections, while saying the companies continued to refuse regulation of the use of writers’ work to train AI to produce new motion-picture content. That establishes a stated union concern at that stage. It does not establish what the September agreement ultimately says about training, consent or any other specific AI provision.

Nor does this dispute require a verdict on whether machines can be creative before the labor questions become intelligible. What work may be used? Under which terms? Who can object? What happens to the person whose paid task changes? These questions survive disagreements about the technology’s capacities.

Coherenceism’s obligation to return what is taken provides a useful ethical test here, not a substitute contract. If a system draws value from people’s work, calling the result innovative does not settle what is owed to them. The answer needs a process in which affected people can contest the terms. Collective bargaining is one such process; its existence does not make every outcome fair by definition.

Relief should not have to perform consent

The September statement thanked members for enduring 146 days of pain and uncertainty. It is reasonable to see the announcement as a source of relief. It would be unreasonable to infer from that relief that every member had already evaluated a deal whose details had not yet been released.

The committee itself preserved the distinction. It promised the agreement materials and meetings before ratification, and explicitly said that a possible return to work during voting would not remove members’ right to decide on approval.

That design gives skepticism a legitimate place inside solidarity. Members need not choose between supporting their negotiators and reading what the negotiators bring back. Agreement is stronger when questions are permitted to remain questions until people have something concrete to examine.

This is also where a coherenceist reading must resist a tidy ending. A quiet workplace is not necessarily a fair one. A resolved negotiation is not evidence that every affected person has been heard. And an approved contract, if approval comes, will still require interpretation and enforcement. These are tests for what follows, not claims about outcomes visible on September 24.

For now, the achievement can be named without borrowing certainty from the future: the parties have reached an agreement in principle, and the union has described how that agreement must pass through written language, leadership decisions and member judgment.

There is enough there to welcome. There is not enough there to declare every disputed protection secured. The next act is less cinematic than a victory speech: make the terms available, keep the distinctions intact, and let the people who carried the strike decide what they can live with.