## Notes from 10 January 2026 [[2026-01-09|← Previous note]] ┃ [[2026-01-11|Next note →]] I've read two posts from [[Rebecca Heywood]] on the incoming Mamdani administration in NYC, plus a piece on OPM's new workforce data site. Heywood lists ten things the Mamdani administration should do to transform service delivery in NYC. While none are novel—integrating backoffice functions (tech, procurement, HR, finance) as strategic partners rather than procedural afterthoughts; [viewing every project through a delivery lens](https://publicsectorjobboard.substack.com/p/what-it-will-take-to-deliver-for-dc2); empowering agency leadership; building iteratively for quick wins—they are essential. The problem, of course, is that this basic integration rarely happens. I know Heywood is fully aware of how often such recommendations go unimplemented. Expectations are high. Over 25,000 people submitted resumes to Mamdani's transition portal within 24 hours. Yet, the civil service system in NYC isn't built to absorb those who want to join. Most positions require exams offered only once every four years (or less), with results taking months or years to process. Hiring managers are restricted to the top three candidates, and the time from offer to start date can drag on for nine months. Furthermore, titles for common tech roles (data scientist, product manager, UX designer) simply don’t exist. Effectively, anyone who hasn't taken a civil service exam in the last four years is locked out. It is a system designed for a different era. While merit protection via examination made sense in the 19th-century context of patronage, today it creates friction and filters out talent. [Heywood notes](https://publicsectorjobboard.substack.com/p/what-it-will-take-to-deliver-for) that NYC has a heavily unionized public workforce but doesn't elaborate much further. In the comments, Gabriel Pincus observes that the system "blocks, rather than advances, its original goal of meritocratic hiring." Although Heywood seems aware that unions can be barriers to modernization, she doesn't pursue this line. Whether strategic or simply outside her focus, the omission is notable. Across different contexts—the UK, US, Germany, Brazil—unions consistently appear as structural obstacles to civil service reform. It remains one of those truths understood but seldom stated directly by progressives in government. Heywood argues for injecting external expertise, framing it primarily around technology roles. However, the argument applies broadly: project management, procurement, and strategic planning all benefit from outside talent. This "delivery first" rhetoric raises an important question: what happens to the bureaucratic apparatus that _doesn't_ "deliver"? Retrain? Reduce? Reallocate? A delivery-centered vision needs an accompanying theory for non-delivery functions and personnel. ## OPM's new workforce data site OPM has launched a redesigned [Federal Workforce Data site](https://data.opm.gov/), which [[Gabe Menchaca]] [notes](https://fedscoop.com/opm-federal-workforce-data-overhaul-fedscope/) exceeded his expectations. The interface is clean, featuring interactive charts filterable by agency and time period, monthly updates, payroll and recruitment data from USAJobs, and downloadable .csv files. One feature stands out specifically: performance ratings by agency. The US federal system uses a five-level evaluation scale. The site displays how employees are distributed across these levels at each agency—including how many are rated as performing at insufficient levels. Having a common evaluation scale across agencies, combined with public data on its distribution, allows anyone to assess whether performance evaluation is meaningful (if 99% are rated "fully successful" or above, the system lacks differentiation); which agencies actually distinguish performance levels; and whether accountability rhetoric corresponds to reality. I would love to see equivalent data for Brazil, but I doubt anyone will ever have access to it. The political sensitivity of exposing how few public servants receive insufficient ratings would be considerable. I’ll write a piece acknowledging how wrong I was if this ever becomes publicly available.