Lectern holds your papers, your marginalia, and the drafts they become — bound into one quiet folio. The assistant is mono-set, ever-cited, never in a hurry.
A paper is read once, marked twice, written about a hundred times. Lectern makes each motion feel like the same motion — a hand moving across paper.
Your library is a folio of bound volumes — papers, working drafts, book chapters, documents. Sort by theme, by venue, by what you've actually finished. The PDF is one click away; so is the chat.
Every paper has a panel of marginalia — your notes, organised by the prompt that occasioned them. Summary, key claims, limitations, anything you invent. A backlinks tab shows where this paper has been cited by your other notes.
@-mentions.The writing room is a manuscript with seven fixed sections — intro · literature · data · method · findings · contribution · misc. Lectern drafts paragraphs from your library, every sentence sourced. You revise; Lectern tightens.
Drag a PDF in. Paste a DOI. Lectern picks up the title, the authors, the venue — then asks where on your shelf the paper belongs.
Themes nest as deeply as your reasoning. Conversations live beside the paper they belong to. The library is a folio of bound volumes — every one of them findable, every one of them indexed for the chat.
The papers that turned applied economics into a question of identification rather than of theory.
Every passage you mark is a question waiting to be asked. Lectern keeps your marginalia bound to the page — your notes by prompt, your highlights with the passage they belong to.
And when you'd rather hear the answer than write it, the same passage is one keystroke from a conversation. The reply is backed by passages Lectern has actually read in your papers — never the rest of the internet, never half-remembered.
Nothing written without a source.
…Contrary to the central prediction of the textbook model, we find no evidence that the rise in New Jersey's minimum wage reduced employment at fast-food restaurants in the state.
They compared employment in fast-food restaurants in New Jersey before and after the April 1992 minimum-wage increase, against a control group in eastern Pennsylvania where the wage was unchanged [p. 772]. Contrary to the textbook prediction that higher wages reduce employment, New Jersey employment rose slightly relative to Pennsylvania [p. 792]. The natural-experiment design reframed forty years of received wisdom about the labour market — and seeded what Angrist later called the credibility revolution.
Drafting is the slow part. Lectern makes it feel less lonely — your library is in the room, your citations are at hand, your sections are laid out like a table of contents.
# Corpus
The reference papers for this essay are @cardkrueger1994's New Jersey study, paired with @banerjeeduflo2011's account of randomised trials in development economics.
The methodological frame is taken from @angristpischke2009 (see @credibility-revolution-essay) — specifically the chapters on instrumental variables and regression discontinuity.
I read the three together; what reads in 1994 as a single, surprising result becomes, by 2019, the dominant grammar of applied empirical economics.
Bring a paper, a question, an unfinished paragraph. Lectern will be ready.