This paper dissects the universe of WTO “special and differential treatment” (SDT) proposals advanced since 2017 and still on the
negotiating table. Using a hand-coded dataset of 88 agreement-specific textual submissions, we classify each provision as “good”
(clearly beneficial to developing-country welfare), “bad” (likely welfare-reducing or protectionist) or “ambiguous” (net impact
conditional on implementation choices). We find that only 29 % of provisions are unambiguously “good”; 34 % are “bad,” often
disguised as technical assistance or transition periods; and 37 % are “ambiguous,” relying on elastic language such as “best
endeavours” or “as appropriate.” Regression analysis shows that ambiguity is greatest where North–South export competition is
intense and smallest where South–South export overlap dominates. Simulations calibrated to 2019 trade flows indicate that
implementing the “good” set would raise developing-country real income by USD 11.4 bn, while the “bad” set would erase USD 7.1
bn of those gains. The paper concludes with a triage rule—quantifiable benefit thresholds, mandatory peer review and automatic
graduation clauses—that could convert today’s mixed bag into a coherent development-oriented SDT architecture.