I'm not sure I understand seasonal adjustment

The unemployment rate rose from 9.8 percent in September to 10.2 percent in October, right? Well, not exactly. The Census Bureau, on behalf of the Bureau of Labor Statistics, surveyed households in September and October. In September it estimated that there were 14.538 million people unemployed and 139.079 million employed for an unemployment rate (u=U/(U+E)) of 9.5 percent. Then it surveyed another group of people in October and estimated that there were 14.547 million unemployed and 139.088 employed, for an unemployment rate of 9.5 percent. The number unemployed rose by 9000 as did the number of employed, and the unemployment rate was unchanged.

But then the BLS seasonally adjusted the data. Given that this is October when presumably some companies start to hire for the Christmas season, given the number of work days in the month, etc., we would ordinarily have seen bigger employment gains. So the numbers were passed through the seasonal adjustment sausage grinder, and voila, we have a decrease in employment of 589,000, an increase in unemployed of 558,000, and an increase in the unemployment rate from 9.8 percent to 10.2 percent.

I get that. But then I compare the unadjusted and adjusted data for October 2009 with October 2008. Comparing Octobers should take care of most of the seasonal factors, no? We're left with adjustments for number of weekdays, weather, whatever else the BLS factors in, which I wouldn't think were very important. But what do we find? According to seasonally unadjusted data, over the year from October 2008 to October 2009 6.455 million fewer people were employed, 5.078 million more were unemployed, and the unemployment rate rose from 6.1 percent to 9.5 percent (an increase of 3.4 percentage points). According to the seasonally adjusted data, we had 6.382 million fewer employed, 5.479 more unemployed, and an increase from 6.6 to 10.2 percent in the unemployment rate (3.6 percentage points).

So in a nutshell, seasonal adjustment subracted 73,000 people from employment and added 401,000 to unemployment, and added 0.2 percentage points to the unemployment rate, from October 2008 to October 2009. I wonder why? And I wonder if this should make me discount the sharp rise in the unemployment rate in October 2009 as somewhat a figment of the BLS's seasonal adjustment techniques?