I like to think about communication in four tiers from best to worst

  In-person: all context clues available
  Video: screen acts as barrier; latency inhibits rapport
  Voice: all visual information missing
  Text: all visual and tonal information missing

The article's advice rings more and more true the further down this list you go