"Help Decode Secret Civil War Telegrams": How reliable and helpful are crowdsourced efforts?

For those of us who are detail-minded, enjoy history and puzzles, and can read 1860s cursive, this is really fascinating. It wasn’t easy but I just now transcribed telegrams by Ambrose Burnside and Andrew Johnson at this site:

My factual question, though, which I don’t see answered in the article that I excerpted below, or in the actual website for the project, is about the usefulness and reliability of crowdsourcing in general, and of the volunteer transcriptions specifically. I know that I’m fairly good at puzzles and at deciphering cursive, and given my reading on the Civil War can make educated guesses–the Burnside telegram that I transcribed wasn’t that legible, but because I know that Burnside was one of the generals, I was able to make out the name. But what happens when a volunteer armed with good will but little else wades in and comes up with wildly inaccurate transcriptions? Wouldn’t that be more hindrance than help? Or are we looking at a self-selecting population with people not volunteering if they know they can’t do it?
You Can Help Decode Thousands of Top Secret Civil War Telegrams
Read more: http://www.smithsonianmag.com/smart-news/you-can-help-decode-thousands-top-secret-civil-war-telegrams-180959561/#WYlBe7U40WIzDl3B.99

Volunteers will transcribe and tease out the messages of of nearly 16,000 communiques
By Erin Blakemore

JUNE 28, 2016

When President Abraham Lincoln wanted to correspond with his generals and cabinet on top-secret Civil War business, he knew he could trust the United States Military Telegraph Corps. Using the era’s most cutting-edge technology, the group transmitted tens of thousands of telegrams that helped dictate the very course of the war. But what exactly did those telegrams say? That’s long been unclear—and now a new project wants you to help find out.

Decoding the Civil War is looking for citizen volunteers to help transcribe nearly 16,000 Union Army telegrams that Thomas T. Eckert, who headed up the War Department’s Civil War telegraph program, saved.

After the war, Eckert never got rid of the top-secret telegrams or the cipher books, the Huntington Library notes. Now, members of the public can view digitized copies of both ciphers and coded messages, crack and transcribe them, or try their hand at uncoded telegrams. The hope is to help present a new view of the Civil War—one that recorded progress not just in terms of North and South, slave and free, but dot and dash.

Also I need to add that when transcribing I could be overconfident and mess up without realizing it!

Transcribing old census records was (and is) done by volunteers. The problem here is bad handwriting and obscure abbreviations, but is broadly similar. No one individual’s transcript has been accepted without independent verification, so accuracy is pretty high, as it has to be.

I’ve done a few of those telegrams, and intend to do more. And boy, let me tell you – the near-death of cursive in our society has seeped into my brain so that I struggle sometimes with the handwriting. I would say that I’m probably 80, maybe 85% confident in what I input into the system. I suppose that’s better than not having someone do it at all, but I also don’t know what the quality control process is after I submit something.

I would add that I’ve participated in other projects like this to a limited extent. There was one about identifying animals in Africa that were taken by an automated camera – that was fairly tough, maybe I had about 75% confidence in my abilities. Then there was one to identify various celestial objects. I had pretty low confidence that I was doing that one right, so I stopped after not too long, especially when it became clear to me that I wasn’t getting the hang of it.

as bob++ implies, this is a common problem in census and other genealogical records.

The way I’ve seen it done (at least for census records) is that any given page is given to at least two different transcribers. If their results agree, then that page is indexed with those results. If they disagree, it’s sent to a third indexer to see which is correct. If someone later finds an error in the index, they mark it as such, and it goes to yet another indexer to verify the change.

It’s not perfect, because ye olde cursive by yon lowly paid scribes can be pretty bad, not to mention the scans are often terrible, but it produces some good results.

We did lots of this in the old census record.

This was in a prison, so the reading standards are not exactly good, however you develop an eye for it.

Some professions would catch you out, things like ‘scutcher’ ‘arkwright’ and ‘fellmonger’ there were lots of others - and the words would be very unfamiliar - obviously if its an archaic word it’s so much harder to identify.

I think those old work roles are perhaps one of the best glimpses into early Edwardian life.

Thanks everyone, good points. I realize that part of my comfort with cursive comes from my longtime interest in fountain pens and calligraphy. And I’m still unhappy but no longer surprised to meet people in their teens and twenties who can’t read cursive at all.

As for census records I’ve seen enough wrong transcriptions to know how challenging they must be.
About work roles that are now unrecognizable, there are common enough family names that refer to archaic occupations too like Tucker and Fuller.

Anyone my age (50’s) or older should be able to read most cursive script. I’m quite comfortable with it if it’s written legibly.

I’ll set aside time and give this project a try.

The zooniverse people have a lot of these sorts of crowd sourced image recognition projects going. I did a little work on classifying galaxies in deep field astronomy pictures.

I ran across the word incepantly in a telegraph. I took me a minute to realize if was inceſsantly. The use of a long s was archaic even back in 1863 but somebody was still clinging to it.

I’m in my 30s and can read Cursive - we were taught it at school in New Zealand, where I’m increasingly convinced I somehow ended up accidentally being taught in the 1950s somehow. :stuck_out_tongue:

Being able to read cursive has been extremely helpful for my interest in history - it means I can read old records, census data, forms, etc.

More interestingly, there are still a few people about who write in cursive and from time to time someone will come to me at work or wherever and say something like “Hey, Martini, you can read old-timey handwriting, right? I’ve got this letter from my great-grandmother; can you read it for me?”

Given telegrams were charged by the letter in some places, it’s also possible someone was trying to save some money too… :slight_smile: