The 1950 Census Follies!

As I indicated in The 1950 Census Blues, in my previous post, I looked forward to looking for and finding myself in the census.  1950 was also the first appearance in a census for my late brother, who passed away in 1996.  So both of us in our first census was special to me.

 Then the "fun" began.  Looking for was one thing.  Finding was something else entirely.

My mother and father never owned a house.  While he was alive, he always rented.  Part of that habit was due to bouncing from job to job.  He had some engineering jobs, opportunities to use his degree in engineering from the U.S. Naval Academy.  He had other jobs, as well.  But the movements of the family looked like migration on steriods -- or on meth!  Between the end of World War II and my father's dischage from the Navy, and 1954 when he died, my dad had the family bouncing between California and Florida nearly every year.  And in between, the family didn't stay put within the locality, either.  Tracking all this down was a real detective story, and this census is only part of that picture, I have developed a keen empathy with ping-pong balls!  I'm also amazed that my mother didn't divorce him for making her pack up and unpack at least once a year, if not more!

So frequent relocation is part of this picture of why it was so doggone difficult to find my family in the 1950 census!

The other part has been both frustrating and outrageously entertaining!  And both of these diametrically opposed concepts have sprung from the National Archives' experiment in allowing an AI to do the first run of indexing the census. Our younger daughter correctly insists that AI is not the proper term.  The proper term, she says, is "machine learing," because the thing may be long on artificial, but it sure is short on intelligence!

The machine learning is mangling names most hilariously. I have told relatives about the manglings.  A cousin working on a Master of Fine Arts degree is writing a novel, and she took one of the name manglings for a character in her novel.  And oh, did it ever do a number on my father's name.  My father's name was Arden Packard.  

The machine rendered -- and believe me, rendered is the exact right word -- his name as . . . wait for it . . .  Urden Pucbard.

I laughed for a solid hour, partly at the relief of having finally found my family -- and me and my brother -- in the census, but mostly at the utter absurdity of Urden Pucbard!

Also entertaining is the machine's mangling my brother's name, Ned, as Nee.  It put my brother's son (and me) immediately in mind of that immortal Monty Python schtick: the Knights who say Ni!

But unfortunately, there is a serious side to the entertainment, sad to say.  Not only is the machine learning or AI or whatever you choose to call it mangling names to an absolutely absurd degree, it is also: 

• skipping individuals 
• skipping entire households
• conflating information on two different lines

 Thus, there are some people who, unless the entire census is transcribed by us human beings who are doing so, will have little chance of finding their family in the 1950 census.

 

Accuracy counts.