Roadmap
An honest, public account of where ScienceEcosystem stands — what's built, what's actively in progress, and what depends on what. Built by one person, in the open, updated as work happens.
Stage 1 — Be useful today Substantially done
The foundation: paper discovery, researcher profiles, topic pages, a personal library, and the tools needed to run a platform that stores user data responsibly.
- Done: Paper pages (metadata, abstracts, citations, figures, citation graph, open access status), author profiles, topic pages, library with Zotero sync and PDF storage, full-text/DOI/boolean search, Chrome extension (Firefox submitted), admin stats dashboard.
- Done: CSRF protection and rate limiting on all endpoints that write data.
- Done: Terms of service and a CONTRIBUTING guide.
- This page. The last item on the Stage 1 list.
Stage 1 unblocked Stage 2 as of 2026-06-20 — the security baseline (CSRF + rate limiting) had to be in place first for a platform handling user accounts.
Stage 2 — Making the conversation visible In progress
Today, a citation count tells you a paper was cited 47 times. It doesn't tell you how — whether those papers built on it, contested it, corrected it, or just mentioned it in passing. Stage 2 makes that visible, and gives cited authors a voice in their own citation record.
- Done — "Who cited me": researcher profiles list papers that cite their work, sourced from OpenAlex.
- Done — Citation context: paper pages show the actual sentences from citing papers where a work is discussed, not just a count. Sourced from the Semantic Scholar API, with an OpenAlex-abstract fallback when Semantic Scholar doesn't have a citing paper indexed yet.
- Built, not yet live — Citation stance classification: automatic labelling of each citation context as Supports, Challenges, Corrects, Extends, Uses method, or Background, using the Claude API. Currently held behind a kill switch pending cost and abuse hardening before it's exposed to real traffic.
- Done — Author response: paper owners can add a note responding to how their work has been cited, visible to all readers.
Current bottleneck: the Semantic Scholar citations endpoint (/paper/{id}/citations) intermittently rate-limits even with a valid, correctly-provisioned API key — a stricter quota on that specific endpoint than on basic lookups. Being retested now with a longer wait between attempts. This is what's holding Stage 2 from being fully done and Stage 3 from starting for real, since Stage 3 depends on citation-context data accumulating at scale.
Stage 3 — Topic pages as field records Depends on Stage 2 data
Once citation stance data exists at scale, topic pages can move beyond "here are the top papers" to something closer to a live field record: where a research area has reached consensus, where it's contested, and where the open questions actually are.
- Aggregate citation stances (Supports / Challenges / Corrects / Extends) at the topic level, not just per-paper.
- Surface contested claims and open questions algorithmically, grounded in real citation behaviour rather than editorial judgement.
- Estimated timeline: 6–12 months after Stage 2 is complete, depending on how much citation-context data accumulates in the meantime.
Not started for real yet — needs Stage 2 stable and generating stance data at meaningful volume first.
Further out
- Medium-term: claim extraction from abstracts and results sections, claim-to-claim linking across papers, a community annotation layer with governance rules.
- Long-term: a full claim graph across the indexed literature, and a "living paper" format — new papers published directly into the graph as structured, linked claim sets rather than static PDFs.
These depend on funding and contributors. The full strategic narrative is on the business plan page.
Follow along or get involved
ScienceEcosystem is built by one person, in the open. The code is on GitHub — issues, pull requests, and ideas are welcome.
Email: info@scienceecosystem.org