Was brought to you via your Coral Project post. There are good ideas in this essay. Some quick thoughts:
- I am dubious about relying heavily on mobile. Most mobile setups favor TLDR because it’s too hard to write and annotate without a full keyboard. Screens work fine, mobile keyboards don’t. A swipe is fine for a “like” or a vote, but I for one would never author a message like the one I’m typing now on my 6P, Samsung tablet or Kindle Fire.
- Machine learning without robust natural language processing systems and ontologies have limited usefulness.
- The piece speaks of hypothesis-oriented discovery: +1 for that.
- I’m wary of sentiment analysis. It took years to ax the Facebook “like” in favor of more nuance. Think “likes” to eulogies and reports of deaths in the family.
- Walls erected between academic publishing and “news” (in Bibtex, a “news story” is quietly deprecated vs. “journal article”) are tall.
- A lesson from Twitter is that there is value in short summaries, and speedy link-outs. This is very different from providing context in depth. The two must play together somehow, but I haven’t seen this working anywhere yet.