IssueTracker

What steps will reproduce the problem?
Run the following query in bigquery.cloud.google.com with 'Use Legacy SQL' not checked. This query is using public datasets.
```
WITH AllNames AS
(
  SELECT DISTINCT name
  FROM `bigquery-public-data.usa_names.usa_1910_2013`
),
HackerNewsAuthors AS
(
  SELECT DISTINCT author
  FROM `bigquery-public-data.hacker_news.comments`
)

SELECT * FROM AllNames
WHERE name NOT IN (SELECT author from HackerNewsAuthors)
```

What is the expected output? What do you see instead?
I expect to see approximately 30000 rows, but instead I get 0 rows.


What version of the product are you using? On what operating system?
I'm using https://bigquery.cloud.google.com/ on Chrome Version 54.0.2840.71 (64-bit) on Ubuntu 16.10.

Please provide any additional information below.


The following query has more promising results (~30000 rows).
```
WITH AllNames AS
(
  SELECT DISTINCT name
  FROM `bigquery-public-data.usa_names.usa_1910_2013`
),
HackerNewsAuthors AS
(
  SELECT DISTINCT author
  FROM `bigquery-public-data.hacker_news.comments`
)

SELECT * FROM AllNames
WHERE name NOT IN (SELECT author from HackerNewsAuthors WHERE author <> 'supercalifragilisticexpialidocious')
```

The following query in Legacy SQL also has promising results (~30000 rows):
```
SELECT * FROM
(
  SELECT name
  FROM [bigquery-public-data:usa_names.usa_1910_2013]
  GROUP BY name
)
WHERE
name NOT IN
(
  SELECT author
  FROM [bigquery-public-data:hacker_news.comments]
  GROUP BY author
)
```

Note: the author column in bigquery-public-data:hacker_news.comments has null values

Job ID  pandas-140401:bquijob_6b047e13_158a0d8b668
Creation Time   Nov 26, 2016, 8:35:07 AM
Start Time      Nov 26, 2016, 8:35:08 AM
End Time        Nov 26, 2016, 8:35:11 AM
Bytes Processed 119 MB
Bytes Billed    120 MB
Billing Tier    1
Destination Table       pandas-140401:_ed72b339b3c5a29a8fa0bcb848b39cb9daa256e4.anon64986255_1b2d_4626_a878_48b35668af3f
Use Legacy SQL  false