ancestry 1911 update

[locksmith]

I think the handwriting has improved, but there are more transcription errors that seem inexplicable. The Eltham, Fulham typ of errors are not in isolation. In previous census years you could have had pages of almost illegible writing from a single enumerator but a remarkably successful transcription. These transcriptions do appear to have been done in haste, possibly automatically and probably by those unfamiliar with British names/nameplaces.

When the Findmypast 1911 census first came out, transcriptions were pretty poor as well but at least when a correction is accepted the transcription is changed. Ancestry's transcription errors remain with an alternative. Easier than correcting I suppose.

Simon

[Alexander.]

I don't think the Eltham=Ham, Fulham=Ham, Ilkeston=Keston, etc. is an error made on the part of the transcribers. Since it is so widespread, it looks more like to me an error when Ancestry compiled the database. Assuming the original data is correct and still exists, I would think it may be an easy thing to fix.

The 1911 census is particularly difficult for transcribers because each household is in a different hand. At least on previous censuses, as you transcribe a town or village, you can get used to the way the letters are formed and see trends. Can't do that on the 1911!

Really the transcription errors are no more appalling than those found on FindMyPast for this census. I don't quite understand why people seem to expect Ancestry to invest more time and money into producing a better transcription than other commercial sites? Poor Ancestry really can't win, no matter what they do. Either people are complaining why certain records are not available yet, or complaining that they did the work too quickly.

Besides, if everything was transcribed perfectly the first time, that would take half the fun out of our searches... 

[Koromo]

I don't think the Eltham=Ham, Fulham=Ham, Ilkeston=Keston, etc. is an error made on the part of the transcribers. Since it is so widespread, it looks more like to me an error when Ancestry compiled the database. Assuming the original data is correct and still exists, I would think it may be an easy thing to fix.

That was my first thought (hope!) when I saw it was the last three letters of a placename. But Wall for Cornwall and Wark for Southwark are the last four letters and how do you explain Woolwich=Wool, the first four letters, and Southampton=Ham, the middle letters?

I really can't agree that ordinary human errors in transcriptions can make it "fun" — understandable, yes, but absurd wholesale errors are unacceptable. What about the people who can only afford to buy the occasional credits which are used up very quickly by (avoidable) ridiculous transcriptions?

And by the way, Ancestry is not poor!

[Carmela]

I. These transcriptions do appear to have been done in haste, possibly automatically and probably by those unfamiliar with British names/nameplaces.

According to  a registration statement filed by Ancestry.com Inc  with the US Securities and Exchange Commission in 2009, in preparation for going public on the US Stock Market, the majority of Ancestry's data transcription was then done by a company called  Formax, based in Beijing, China. Don't know if they are still using the services of Formax. Could explain the mistranscriptions.
 
Carmela

[Nick29]

The 1911 census is particularly difficult for transcribers because each household is in a different hand. At least on previous censuses, as you transcribe a town or village, you can get used to the way the letters are formed and see trends. Can't do that on the 1911!

You could equally argue that the 1911 census should be the most accurate transcription of all, because it has only been transcribed once, and not twice, like the 1901 and previous censuses.

[Nick29]

I. These transcriptions do appear to have been done in haste, possibly automatically and probably by those unfamiliar with British names/nameplaces.

According to  a registration statement filed by Ancestry.com Inc  with the US Securities and Exchange Commission in 2009, in preparation for going public on the US Stock Market, the majority of Ancestry's data transcription was then done by a company called  Formax, based in Beijing, China. Don't know if they are still using the services of Formax. Could explain the mistranscriptions.
 
Carmela

Sorry, I disagree.  As Stanmapstone has said, transcribers are required to type what they see, and sometimes local knowledge can be a hinderance in doing this.  The two places  Eltham and Elham in Kent have already been mentioned in this thread.  I went to school in Eltham, but I'd never heard of Elham until I started researching my tree, and in the early days I frequently mis-read Elham as Eltham.  If I'd been paid to do any transcribing at that time, I would have happily mis-transcribed Elham as Eltham, because of my 'local knowledge' 

Ancestry's continual mistranscriptions of very clear place names which contain the letters HAM are not the work of Chinese transcribers (whom, allegedly also transcribed BrightSolid's 1911 Census).  I can accept Freckneale's suggestion that this was the work of those who compiled the database and indexing, and not the transcribers, who seemed to have little trouble transcribing other parts of the census.

[Ray T]

I'm currently wondering what Consumer Direct/Trading Standards would make of this. It appears that we're all paying for something which isn't fit for purpose.

[ReadyDale]

I don't think the Eltham=Ham, Fulham=Ham, Ilkeston=Keston, etc. is an error made on the part of the transcribers. Since it is so widespread, it looks more like to me an error when Ancestry compiled the database. Assuming the original data is correct and still exists, I would think it may be an easy thing to fix.

The 1911 census is particularly difficult for transcribers because each household is in a different hand. At least on previous censuses, as you transcribe a town or village, you can get used to the way the letters are formed and see trends. Can't do that on the 1911!

Really the transcription errors are no more appalling than those found on FindMyPast for this census. I don't quite understand why people seem to expect Ancestry to invest more time and money into producing a better transcription than other commercial sites? Poor Ancestry really can't win, no matter what they do. Either people are complaining why certain records are not available yet, or complaining that they did the work too quickly.

Besides, if everything was transcribed perfectly the first time, that would take half the fun out of our searches... 

The Fulham=Ham type errors are the most laughable of the problems, but yes are probably down to some sort of glitch, so should be easily correctable. However, they are not the only errors. In the seven days since the transcriptions went online, I have submitted well over fifty corrections (excluding the Ham errors). It probably would have been more but unfortunately I have to work too . I think we all understand the writing issues, but in the vast majority of the errors I reported there was no problem with the writing. In fact it was virtually impossible to work out how they got to the words that appeared in the transcriptions!!!
My understanding of transcription projects (like FreeBMD, etc), is that everything is transcribed twice and the results compared, with differences flagged up. It would be interesting to know if this method was used for Ancestry's 1911. It certainly looks as though they got just one person to do it and used the results. If they did adopt the multiple pass approach, then they should be seriously looking at whether they got a good deal from their outsourcing company!!

[Koromo]

Ancestry have replied again on the blog:

http://blogs.ancestry.com/uk/2011/12/08/1911-census-%E2%80%93-millions-more-searchable-records/#comments

Most of the specific issues reported have centred around problems with the birthplace field in a minority of the records. Some of you have already correctly identified that this has not been caused by a transcription error but a glitch in how our systems are presenting the data. While we know what the problem is, developing and implementing a solution is not always quick or easy but we assure you that we are working hard to fix it."

There is hope!