Description
This talk is about the verifiability crisis in science, politics, and technology, and how methods developed to address verifiablity in statistical publishing can be adapted to other areas such as technical writing.
- In 2015, a group of researchers set out to replicate the results obtained in a set of 100 published psychology research papers. It turned out that only 29 out of the 100 could be replicated.
- In the 2016 elections, so-called "fake news"--fabricated news published on venues such as Facebook--influenced many voters in their choice of candidates.
- In the field of technical writing, it is not uncommon to discover that technical documentation is inaccurate. Although the docs might have been accurate when first published, the underlying technology evolved rendering the docs obsolete.
Academic, public-policy, and business decisions are sometimes made after accepting published (mis)information at face value. The costs can beconsiderable.
The information often arrives in the form of an electronic document, such as a web page or PDF file. To verify the information, someone has to "fact check" it. But what if the document could, in some sense, "speakfor itself" and demonstrate the veracity of what it is saying?
In response to the verifiability crisis in science, statisticians havedeveloped technology to create "intelligent documents" that not onlyreport the results of a statistical analysis, but can also re-create theanalysis itself.