X'ing submitted docs: Dos and Don'ts

[jvwong]

Description

Improvements can be made to the structure and content of newly submitted documents shared to Twitter. Ideally, we would produce more compelling content on a per-document basis ('single-user mode') by leveraging the unique and valuable information only we have access to via Biofactoid. Ultimately, we'd like to achieve some positive feedback: increasing the chances of views, re-tweets, indexing (Google) prompts others to contribute.

Some Tweet "Dos and Don'ts" to use as a guide:

Dos

• Tag
  • With some prioritization and limits (number)
  • Authors with unambiguous account names
    • sourced from ORCID (Add Twitter Mentions for Factoid Article Authors from ORCID  #1017)
  • Institutions, labs
• Display a compelling image
• Hashtag
  • topics that are unambiguous, popular and useful
• Text
  • Narrative
  • Summary
    • GPT
    • Semantic scholar TLDR feature
• Links
  • Directly to the paper
    • via ID that can be normalized
      • including: DOI; PubMed ID; ISBN; arXiv IDs; ClinicalTrials.gov records
      • Altmetric > Sources of Attention
      • CrossRef > Event Data Deprecated

Don'ts

• Split Tweet content into multiple posts e.g. thread
• Tag/Hashtag
  • Everything and everyone possible, as this detracts from content and looks like spam

Specification

To illustrate how we might apply these rules, I'll use a paper from eLife for which a hand-crafted Tweet was created. This paper is rather broad in scope (not just about an interaction), contains complex details, a broad range of experimental techniques and subtle conclusions.

• Arian Mansur et al. Dynamic regulation of inter-organelle communication by ubiquitylation controls skeletal muscle development and disease onset. July 11, 2023. 10.7554/eLife.81966
• Biofactoid document

Mockup

The mockup below is from a 'Features' series whose intention was to highlight the official recommendation by eLife in June.

Details

Some breakdown of the above example:

• Tag
  • journal directly (@elife) - could be automated?
  • Senior author (@@LoweLabMSKCC) - manual
• Hastag
  • main disease (#hepatocellular)
• Link
  • DOI, biofactoid doc in same tweet
• Key image
  • figure - manual

[jvwong]

NEWS: X is changing how news links show up on the timeline.

It will strip out the headline/text so links display only an article’s lead image.

• issues
  • text is elided (this stream of template interactions text was never great anyway)
  • title/caption is gone