IAB Tech Lab's Redefining Media Types would classify OpenRTB video impressions using fields sellers already send

Per the dev.to explainer, RMT works as a reducer over the bid request. A classifier takes the JSON an exchange already accepts, walks a fixed table, and emits eight yes/no answers plus one macro bucket ^[8]. None of it needs a new endpoint ^[15]. We think building on fields sellers already send is the right design for adoption, because it asks nothing new of the transport.

Sound state is where the reduction gets awkward. OpenRTB defines video.playbackmethod as a list of the playback methods that could be used for the placement ^[9]. Values 1 and 5 start the ad with sound on. Values 2 and 6 imply sound off by default ^[9]. RMT treats the field as one factual claim about one impression ^[9]. A request carrying playbackmethod: [1, 2] is valid and asserts both, and the post says the industry has to agree it cannot mean both ^[15].

Skip spreads the same problem across three fields. The attribute reduces video.skip, skipmin and skipafter, and OpenRTB only defines the offsets for requests that set skip to 1 ^[10]. A request that sends skipafter but never says the ad is skippable still passes as JSON. Nothing defines what it commits the seller to ^[10]. The post calls any skip bit derived from that trio "guesswork dressed as measurement" ^[10].

The remaining attributes are inferred from fields that were never written as declarations. Addressability is read off device.ifa, device.lmt and user.eids ^[13]. For measurability, the proxy is whether video.api includes OMID, enum value 7, along with any metric objects on the imp ^[13]. Device class reads device.devicetype alongside the app, site or DOOH context ^[12]. Ad format reads plcmt, linearity and mtype plus the CTV Ad Portfolio extensions ^[11]. Pause, overlay and squeezeback formats only make sense when plcmt and the related AdCOM enums match what the player can render ^[11].

That dependence on plcmt has a history. The IAB deprecated video.placement in 2022 and introduced video.plcmt with tighter definitions ^[16]. Three years later, according to a migration retrospective the post cites, plenty of traffic still carries contradictory values in both fields, or neither ^[16].

The open comment window the post points to belongs to Programmatic Standard Practices v1. It opened on September 16 and runs 30 days ^[1][19]. The post places RMT in a different Tech Lab thread ^[3]. PSP v1 says its goal is agreement on how the existing standards should work in live trading, with no new pipes built ^[2]. In our view, the field agreements RMT depends on, for playbackmethod, plcmt and skip offsets, are exactly that kind of behaviour ^[15]. The post does not say whether RMT itself is part of that comment period.