I'm currently working on a query that should return a subset of a CartoDB table (i.e. a new table) sorted by proximity to a given point. I want to display labels on the map corresponding to the closest, second closest, etc. and thought to capture that by using the PostgreSQL row_number() method in a new column:

SELECT
    *,
    ST_Distance(
        ST_GeomFromText('Point(-73.95623080000001 40.6738101)', 4326)::geography,
        the_geom::geography
    ) / 1609 AS dist,
    row_number() OVER (ORDER BY dist) as rownum
FROM locations
WHERE ST_Intersects(
   ST_GeomFromText(
      'Point(-73.95623080000001 40.6738101)', 4326
   ),
   the_geom
)
ORDER BY dist ASC

However, when I try this, CartoDB/PostgreSQL returns the following error:

Error: column "dist" does not exist

Any suggestions on a better approach or something I'm missing?

You CANT use a field calculated on the same level.

SELECT (x1-x2)^2 + (y1-x2)^2  as dist, dist * 1.6 as miles
                                       ^^^^^
                                    undefined

So you create a subquery.

SELECT dist * 1.6 as miles
FROM ( SELECT (x1-x2)^2 + (y1-x2)^2  as dist
       FROM table
     ) T

more