Some of the questionable discrepancies may be explained away with the timing of adjustments as mentioned in the article, but not all of them.
Many of you may recall the suspicions when numbers were released prior to the election. The unemployment rate took a timely nose dive. Digging into the numbers we saw the number of new jobs in the employer survey were a small fraction of the number of additional people who said they were working in the household survey.
More recent data show a reversal of that glitch. There were a total of about 300,000 jobs added in the employers survey in November and December, but the count of people working in the household survey did not increase.
So, it does look like BLS fudged numbers prior to election.