Without knowing much about your actual application (i.e. treat the following with utmost skepticism!), my instinct would be to separate the fact of observing a name from the fact of the observation count. I would keep a log table and a summary table like this:

  CREATE TABLE namelog (
    name TEXT
  );

  CREATE TABLE namesummary (
    name TEXT,
    seen int
  );

Then each time you see a name, log it:

  INSERT INTO namelog VALUES (%s)

Then on a regular basis, run a housekeeping/summary process like this:

    CREATE TEMPORARY TABLE prevsummary AS
        SELECT * FROM namesummary;

    TRUNCATE namesummary;

    INSERT INTO namesummary
        SELECT coalesce(a.name, b.name) as name,
               coalesce(a.seen,0) + coalesce(b.seen, 0) as seen
        FROM 
            prevsummary a 
            FULL OUTER JOIN
            (SELECT name, count(*) as seen
            FROM namelog
            GROUP BY name) b
            ON a.name=b.name
    ;

    DELETE FROM namelog;

This of course is grossly space-inefficient, but making the logging process INSERT-only should enable a great degree of concurrency - and you can tune the summary process to run as often as necessary to keep memory/disk usage within your available resources.

If concurrency isn't an issue, then your initial solution is probably fine.