Freeform Breakout 

Possible topics include:

1.  What’s the next killer application for program synthesis?
2.  How can we better cooperate as research community?  Is there tooling we can share?  Datasets?  Should we define benchmarks?
3.  Can we increase developer uptake of our tooling?
4.  How can we escape Clippy’s curse (elaborated below)?
5.  Topic suggestions from the audience.

Combatting the Curse of the Clippy

Clippy was a cutsy paperclip animation intended to teach you how to productively use Microsoft products.  It was the UX interface Microsoft engineers gave to the Lumi?re’s project task predictor.  Lumi?re’s tooling watched a user’s interactions and tried to predict what that user wanted to do.  When it was sufficiently confident, it intervened with a suggestion.  When Lumi?re was deployed, space was tight, so much of its functionality was dropped.  The result was Clippy, a long running train wreck and a UI disaster that culminated in Bill Gates announcing, at the release of Windows XP in 2001, “XP stands for Ex-Paperclip” to thunderous applause.
Making an incorrect suggestion is disruptive as the developer needs to consider and recover from an incorrect suggestion. Of course, increasing accuracy mitigates the problem, by reducing the probability of making incorrect suggestions, but is orthogonal to the Clippy problem.  The Clippy problem requires resolving a paradox:  how to unobtrusively make suggestions that may be incorrect.

This page was last modified on 25 Oct 2017.