Jump to content

User:Sj/OA

From Meta, a Wikimedia project coordination wiki

Reminder: this is a wiki; feel free to edit :) See also: science fraud

We need a proper commons for human discovery. With free and open ways to author, index, discover, review, and read research. The OA movement took a step in this direction, explicitly as a reaction to the aggressive enclosure of public research and the rising access costs imposed by publishers, in the decades where digital access became almost free. But that movement was reactive, not constructive; it had no complete ecosystem and has developed few avenues for building lasting institutions to match crude but effective publishing companies. It also hides the fact that sharing ideas has always been an open part of developing language, individual and collective intelligence, culture, and science. It's like calling speech that doesn't check with a speech-broker to see if you can participate "open listening".

Starting with current channels for publishing finished summaries of research: this is a catalog of approaches to building a supportive ecosystem without current pitfalls, and combining some of the best features of current niches.

Approaches to an generative, open ecosystem[edit]

Transformation & Reformation[edit]

Gotta flip them all!, aka "I Can't Believe It's Not Science/Nature!"

  • a) Switch editorial teams over to community-publishing. Compare journals + preprint servers that move over to PubPub, those that switch to more trad university-hosted publishing (Ex.), and those that spin up their own self-hosted OJS or WP site.
  • b) Provide a tertiary summary that's more tractable than the original. WP style: parallel summaries, like the original third-party abstract services - that provide rich context, doc maps, citation graphs, and links to relevant benchmarks / replications / refutations. For many purposes, more useful than the static snapshots put out by the source [which e.g. are almost never flagged/revised/retracted even for discredited or disproven work].

Subverted: Being acquired lets us transform larger orgs from within!, aka optimism makes it easier to sell

  • Hindawi (OA aggregator) acquired by Wiley in 2021, after founders step down
  • Authorea + its OA tech pipeline acquired by Wiley, growth slowed, main features integrated elsewhere.

Recreation[edit]

The Wikipedia approach. Identify the final goal -- e.g., "a free summary of every notable research paper and discovery, with structured data and discussions about it, that anyone can contribute to" -- and iteratively build tools that support a decent community doing this for millions of papers + experiments a year. Summarizing work published elsewhere, developing a new ecos of reputation and visible usage stats.

The xxx.lanl approach. Drop substantive review, save to filter spam, cranks, and abuse. (arxiv, not vixra) Handle classification, style guide unification, indexing, format conversion. Central community handling this basic filtering.

The PLOS One approach. A single meta-journal, making submissions visible immediately, with reviews accruing along the way to completing at least one peer review process. A combination of preprint, print (after review), and postprint (updates, commentary, replications, extended-universe of supplementary and complementary work). Handling review and submission w/ a central community. (anyone can submit and get a first index, but review and a final index / reviewed badges is central)

The replicator approach. Build an ecosystem around a dynamic equilibrium of replication, revisiting, revision. Encourage well-targeted citations and sourcing, discourage vague and self-cites, avoid any metrics that can be gamed at a single point in time (like raw citation count, unmodified by later discovery of rings of editors / reviewers that sell reputation as a service). Reward negative results, replicable descriptions w/ full source data, replications (+ or -). Reward clear debunking of poor methods or work.

Are there good examples of this in practice?

The pubpub approach. Platform for communities to manage their own publishing networks, workflows, and standards. Open-source, community-led, end-to-end publishing for knowledge communities (docs, art, or data).

The citizen science approach. SETI/Folding/Rosetta@Home, Zooniverse, iNaturalist. Distributed observation + analysis, collectively iterated and published as liviing texture w/ occasional article-shaped snapshots. Thousand-coauthor papers (see also: LIGO and other universe-shaking astro papers)

Repository approaches[edit]

Places to store and get citable identifiers for work of any granularity, or to access + interpret them using shared resources.

Infrastructure: storage + compute + services

  • Open Science Grid - submit locally, run globally
  • European Open Cloud - a marketplace rather than a single coherent cloud. Noisy.
  • ESRI, Tableau: visualization servers (arcgis+) - many groups store data here b/c viz dashboards are the only expected uses/interfaces to it. Non-archival.

Code + data repositories

  • Github, Gitlab, &c
  • Colab, BinderHub - computable notebooks
  • OSF
  • Zenodo (via CERN, global)
  • OPERAS (EU)

Observation repositories

  • ADS (NASA, space)

Chained approaches[edit]

"immutable" -- You keep saying that word! I do not think you know what it means.

Overviews:

Projects:

  • ARTiFACTS (site) - indexing granular elements of research. Anyone who needs but never settled on a nanopub service to get a DOI wants something like this.(1)
  • Atoms.org (paper). "collectively reward scientific contributions, including proposals, papers, replications, datasets, analyses, annotations, editorials, and more"
  • IBFS / Blockchain for Science (site) - Zug hub, Planck institute private chain. Supported ARTiFACTS and Frankl. (Miniconf, 12/21)
  • Dat protocol - proudly not a chain, but a distributed datastore.
  • Decent. Science (Papers: 2019, 2021) - open peer review and author/reviewer reputation. Spun off from P2PModels.
  • EUREKA (token), ScienceMatters discussion
  • Frankl (site) - incentivizing data sharing and successful? replication.
  • IPFS - used by OpScientia and others
  • LabDAO (about) - for big-ticket bio R&D
  • Manubot - logs updates to Bitcoin
  • OADAO - Still in the works...
  • OpScientia DAO (site) - automated knowledge-generation foundries + immutable? archives of discovery
  • Orvium (essay)
  • OSC - Open Science Chain (site)
  • Planck manuscripts (about) - post-hoc funding for ms production
  • ResearchHub | ResearchCoin - About. A social platform for researchers to upload, summarize, discuss articles.
  • VitaDAO (about) - $10M treasury to date to fund vita research, communally-held IP. Aiming for public benefit from public investment?
  • Science Matters (ex), in general. (see also EUREKA above)

Others:

  • Pluto?
  • Journals w/ decentralized autonomous review: proposed but not implemented. WP:Peer review is not unlike that for WP articles, but WikiJournals so far chose to use strong identification and central review

Instances of widely-used chained work:

Parallel approaches[edit]

That promote and encourage this ecosystem / would rather live in it but connect w/ existing systems and metrics and collections:

  • Discovery (that can choose how to sort and filter: e.g., not by manipulable lagging indicators like JIF)
    ResearchRabbit
  • Commentary (can be used to see overlapping badges and notes from different communities)
    Hypothes.is
    PubPeer
  • Identity (supporting federation, limiting sock- + meatpuppets) WP approach: OAuth to Wikipedia links to established nyms Nymity
  • Local reputation + review
    Recensio (stub)
    OpenReview (site) - mainly CS conferences, but general tool. Sponsored by CS&S

Reflections + Landscape[edit]

Market Analysis[edit]

  • SPARC landscape analysis: Overall, Springer/Nature (> $3B each as of 2020)

The "unethical, consolidating" market[edit]

  • Elsevier: Turning the Supertanker (on their pivot from running weapons trade shows, and 35% margins) (Deutsche Bank, 200x)

Universities pushing back[edit]

Universities and grantors remain the largest subscribers to, and funders of, the market. Paying for both publication (as professors need this for career advancement, they pay out of pocket; as grantors want good publications as outcomes, they earmark it or add it to grants; as universities use it for their own statistics, they invest in it) and for access (as libraries need access for faculty they subscribe at rapidly inflating annual rates). In countries where university budgets are largely from the state, federal agencies pick up the tab.

  • Sweden, 2018 (THE); Germany, 2018 -- both refused to renew their Elsevier agreements for lack of acceptable OA options, and lost access for a time
  • UC network, 2020: their eventual Elsevier deal and potential problems with that solution.
  • Unsub data: reflecting how much of current subscription-fee access is already available via other channels as OA.

References, Notes, Commentaries[edit]

Overviews and research[edit]

Directories and maps

Research

Curated notes and lists

Legal notes

  • Security vs Utility - "those who would give up utility to purchase a little security deserve neither"

Social context[edit]

A matrix of open science infrastructure requirements, and blockchain technology characteristics that are fulfilling them. Leible, et al, 2019.

The sociology of the scholarly commons[edit]

Why "paying people for contributions" isn't the bottleneck, is a source of perverse incentives, and has fed the current split of market-motivated + science-motivated members of the ecosystem. And how we got here, after millennia of trade secrets and alchemy.


Publisher monopolies adopting and enclosing OA

It's always a better short-term business to find places to charge rent. Publishers whose founding principle isn't OA or open collaboration generally adopt the most restricted form of access they can, negotiating authors and reviewers down to things like "read access, but no reposting or translating or remixing, data about your use must be ours to resell"

  • In pursuit of open science, open is not enough (Aspesi + Brand, 2020)
  • Wiley: had lots of cash on hand in 2016 (~500M), now down to $100M, but still on a steady acquisition spree for small independent OA successes (< $20M each). Authorea, Knowledge Unlatched, J&J Editorial Services...
  • OA Monopoly tag on OATP