Organizers, their sponsors and partners want to know how much work was done during an event. "Bytes changed" has been a metric for this, but it is imperfect in many ways. "Words added" would be a way to judge how much writing was accomplished, but this metric is problematic.
The task here is to figure out how to include a figure for Words AddedThis metric will be used in (to begin with) the Event Summary (T205561 and T206692 ) and Pages Created (T206058 and T205502) reports..
- **Pages Created and Improved reports**: These reports list is a list of articles created and improved, so the metric will show the net change in words to the given article. Use a minus sign to indicate negative numbers, in the event of negative change (possible for Pages Improved, though noThe metric will appear in different for Pages Created).ms in these two reports:
- **Event SummaryIn Pages Created and Improved report:** heres**: These reports list is a list of articles created and improved, so the metric will countshow the net change in words added during the event—essentially a sum of the figures above.
=Caveats and limitations
- **Where possible: **it's understood that determining what is and isn't a "word" is harder in some scripts than others.to the given article. Use a minus sign to indicate negative numbers, in the event of negative change (possible for Pages Improved, The first release of this will bethough not for those scripts where we have relative confidencePages Created).
- **For scripts that are not amenable**, we will fill these fields with an "n/a," to signify "not available."
- **Standard definitions:** I imagine there are various techniques for defining and deriving what is a "word." Any standard definition is acceptableIn Event Summary reports:** here, the metric will count the net change in words added during the event—essentially a sum of the figures above.
It is understood that this figure may be easier to derive for some languages and scripts than it is for others where, for example, established bytes-to-words or characters-to-words conversion rates may not be readily available. It's also understood that the figure may not be accurate—since it may be difficult to separate wikitext from content, for example.
Various methods have been discussed for arriving at this metric. Each has pros and cons. The task here is to figure out how we might most feasibly derive the metric for the most languages, so long as it doesn't include wikitext codes in the countand to give a sense for the level of effort this would require. Accuracy may not be a paramount value: Organizers I spoke with recently (in a session at WikiConference North America) expressed a willingness to tolerate a high level of inaccuracy in order to get the metric.