Remove references to the popularity score (#8646)

Author: sigurdm

doc/help-scoring.md

@@ -14,8 +14,8 @@ The three dimensions are:
   - **Pub Points**: A new measure of quality. This includes several dimensions
     of quality such as code style, platform support, and maintainability. More
     about this below.
-  - **Popularity**: A measure of how many developers use a package, providing
-    insight into what other developers are using.
+  - **Download count**: A measure of how often a package is downloaded,
+    providing insight into what other developers are using.
 
 ## Likes
 
@@ -38,18 +38,6 @@ On the other hand, for some packages the download count may be relatively high.
 
 The download count is not a perfect metric, and should only be regarded as an indicator of popularity.
 
-## Popularity
-
-Popularity measures the number of apps that depend on a package over the past 60
-days. We show this as a percentile from 100% (among the top 1% most used
-packages) to 0% (the least used package). We are investigating if we can provide
-absolute usage counts in a future version See
-[this](https://github.com/dart-lang/pub-dev/issues/2714) issue.
-
-Although this score is based on actual download counts, it compensates for
-automated tools such as continuous builds that fetch the package on each change
-request.
-
 ## Pub Points
 
 Pub points is pub.dev's measure of quality. Pub points are awarded in six

doc/help-search.md

@@ -71,7 +71,7 @@ combination of the following parameters, listed in order of relative importance:
   - README
   - Documentation comments and identifiers
 
-- Package popularity and like count
+- Package download count and like count
 
 - [Pub points](/help/scoring#pub-points), which are based on factors such as these:
   - Adherence to Dart file conventions

doc/policy.md

@@ -188,15 +188,14 @@ For information about other Google brand features, visit
 
 We display *Top packages* on the pub.dev landing page to help you find packages
 for building your applications.
-Top packages are selected [based on popularity][4] (including number of
-downloads).
+Top packages are selected [based on download counts][4] .
 Search rankings are based on: relevance, engagement and quality.
 These elements are given different importance depending on the search parameters
 (such as ordering).
 To estimate relevance we look into many factors, such as how well the title,
 tags, description and package contents match your search query.
-We estimate engagement by incorporating signals such as popularity
-(based on download counts) and likes given by users.
+We estimate engagement by incorporating signals such as download counts and
+likes given by users.
 Finally, quality is assessed through package analysis (see scoring).
 These recommendations and rankings are not personalized to our users.
 

doc/search.md

@@ -11,7 +11,7 @@ adjusted with the text query match score of the search query.
 The base composite score sorts packages according to::
 
   - 50% from pub score (calculated from `pana`)
-  - 50% from usage metrics from likes and popularity
+  - 50% from usage metrics from likes and download counts
 
 The usage metrics use a non-linear scoring model, where the package above the N-th
 percentile gets N/100 points (e.g. if a package has more likes than the 90% of the
@@ -21,7 +21,7 @@ packages, it will get 0.90 points for likes).
 
 Numerical ordering is used on search requests that do not specify a text query,
 but would like to order the results by `updated`, or `created` time, and also by
-`likes`, `popularity`, or `pub score`. 
+`likes`, `download counts`, or `pub score`. 
 
 These work on the raw values, without any weights or transformation.