The screening rule Schreiber describes works as underwriting. He puts Miami's litigation costs above $1 million before fees, covering experts, depositions and data work [5], and says a contingency practice needs a case worth $10 million or more before that spend makes sense [6]. Costs at a tenth of the floor recovery [2] leave no room for a modest-injury claim whatever its merits, and the award that came out the other end was roughly 24 times that floor [1]. The practical effect is that Tesla's driver-assistance docket is being selected, case by case, for the worst available facts.
The compounding advantage here is methodological, built up trial by trial rather than bankrolled. Schreiber's account of the field is that conventional auto product liability has a playbook for tire and seatback failures, with recurring experts and testing already done, while autonomous-vehicle work had none until last year [10]. Software leaves no witness mark to photograph [11]. His answer is to argue the failure from disagreement rather than from code: if three cameras read the same object three ways, he says, he can show a jury that none of them agreed, without Tesla's source code [12]. Each trial that runs that argument leaves transcripts and qualified experts behind it, which are the inputs that make the next case cheaper. Whether his costs are actually falling is an inference worth testing, not a claim he has made himself.
The doctrinal fight in Miami was about the length of the relevant window, not about the driver. Schreiber concedes there is typically an at-fault driver in these cases and says that should not absolve Tesla [13]. By his account Tesla's frame was the last three seconds [14], and his own was a set of decisions years earlier to overstate and oversell the technology until drivers over-relied on it [15]. For anyone shipping driver assistance, that is a relocation of exposure: the discoverable artefact stops being a torque spec and becomes the marketing record.
This is one lawyer describing his own call log, attached to one win in three trials [4] that is currently on appeal [3]. That is fair on the numerator, and Business Insider notes that Tesla and its attorney did not respond to requests for comment [4], so the frame on offer is the plaintiff's. Harder to discount is the fixed cost he has taken on: a firm does not put a named review meeting on the calendar every week [7] for a docket it expects to close. He dates his first Autopilot calls to around 2021 [16], four years before the verdict [3].
The cost floor and the time window are the same fact viewed twice. The expensive half of Miami was proving the years-before story, not the seconds-before one, so an appellate ruling that confines admissible evidence to the moments around impact would not merely retire a theory. It would raise the price of entry and thin the pipeline behind it, and that is the mechanism by which most of those monthly calls stay calls [8].