I have noticed a number of "speculative" e-mails on this topic. Recently, there was a call-in show on this issue on public radio featuring two members of Congress. The following is what I understand is going on based on this two-hour discussion, news reports, and past knowledge.
No one, not the Bureau of Census or Congress, is proposing an "all statistical" census for the year 2000. The issue is whether, after the population is counted by the normal method (mail questionnaire, plus some follow-up) the Census Bureau will be allowed to use statistical sampling/counting methods to determine how many people were missed and thus, supposedly, get a more accurate count.
The decennial census, as we all know, is required by the Constitution for the purpose of determining representation in Congress. There is no requirement in the Constitution that every person be listed nor is there any method of doing the enumeration specified. As most genealogists know, from 1790 through 1840, the enumeration was accomplished by listing the heads of households and then giving the number of individuals within that household in various age categories. From 1850 on, each person has been listed, with progressively more information about the family and its members gathered by the enumerators. The added demographic and economic information was sought both to provide additional information to the government about the population and as an aid to businesses.
The decennial population count has become extremely important in a number of ways not relating to representation in Congress. For instance, there are numerous programs involving federal assistance that are based on population.
For the past several censuses, largely in response to complaints by cities that the mail questionnaire method results in an undercount of minorities or underprivileged persons, the Census Bureau has used various sampling techniques to adjust the final count.
At issue now is whether these statistical methods can be used in determining the final count for the 2000 census. As you might guess, the Democrats feel they benefit most by the use of these techniques because they result in more representation for those whom they feel traditionally vote for their candidates. The Republicans oppose use of these techniques for the same reason and have succeeded in attaching a provision to a supplemental appropriations bill now before Congress which will prohibit the use of statistical sampling in deterring the final count. The president has said he will veto the bill if it contains this provision.
There are some interesting sidelights. For example, during at least the past three censuses, sampling has been used in that one out of every six households gets the "long" form, which asks dozens of questions about the household. The number of TVs, refrigerators, etc. in the U.S. are extrapolated from this sample. There is some question as to whether this sampling process would be prohibited under the proposal being considered. Since this demographic information is extremely valuable to business, obtaining it would then require everyone to get the long form-with the possible result that even fewer people would return the questionnaires, especially by those in groups that are felt to be traditionally undercounted.
The Republicans proposal anticipates that additional funds will be appropriated to hire additional enumerators to do the follow-up counts in certain areas. The Democrats say the cost of this is estimated to be $200 million and cite scientific data which indicates the resulting count from this would be less accurate than it would if statistical sampling methods were used.
Bottom line for genealogists in the year 2072: No matter what the outcome of the current debate, the population schedules for the 2000 census will probably be no more accurate/inaccurate than they have been for the last 200 years!
Further, as I have stated before, the reason past population schedules have been so important to genealogists is the lack of other data, in particular, vital records. For genealogists 75 years from now, there will be such a large amount of other data about persons now living that the census will not be an important tool. In fact, it likely will be consulted only in desperation.