Friday, 22 April 2016

When the Digital Age Makes You Scream



Last year, I wrote about the woes of using findmypast for newspaper searches at When the Digital Age Hinders. I now want to follow that up with an evaluation of another newspaper search engine — it’s starting to look like reliable searches are a thing of all ours pasts.

In the findmypast case, there was no visible attempt to address my issues. I made the company aware of the article but there was no comment and no requests for more information, merely a half-hearted attempt to blame the British Newspaper Archive (BNA) for some of the issues. Having tried to use it today (22 Apr 2016), the problems are still there, and even seem to be more intrusive than before.

Punishment of Sisyphus
Figure 1 - Punishment of Sisyphus.[1]

Also today, I needed to perform some searches of The Gazette (UK). Although I had used this for previous research, I recalled that they had redesigned some aspect of their interface; maybe it had been improved.

Unfortunately, I couldn’t get my searches to yield anything resembling what I asked for. I clicked on the little link entitled “Help with search”, just above the text-search field. Rather than giving some textual instructions, though, this directed me to a private — and hence inaccessible — YouTube video called "searching the Gazette" at https://www.youtube.com/watch?v=yJI--4jBhyM. Being part-way through your research, and then being expected to stop and watch some video, is not a helpful plan. After meandering around the site, I found a different link for the same video at https://www.thegazette.co.uk/videos/searching-the-gazette. This was nearly 3 minutes long and eventually informed me that putting a phrase in quotation marks would return only notices containing that exact phrase. But that’s what I had been doing!

I was actually searching for the phrase “Kirk of Nottingham” in all notice types between 1840 and 1860, and it gave me 17 hits. However — yes, you’ve guessed — I couldn’t find my exact phrase in any of them. It took me a while but I did find that each of them contained “Kirk of” but not “Kirk of Nottingham”. Now I wouldn’t insult the designers of this system by suggesting that they only honour the first two words in any exact search phrase, so it must be some inbuilt relaxation of the search terms.

The problem is that it doesn’t deliver what they said it would. They said it would only deliver notices that contain the exact phrase. I have seen several form-fill search engines that describe a field as “exact phrase” and which is probably being relaxed later by some component of the search software. I did notice that the BNA search fields have an “Exact search” tick-box but I couldn’t tell whether it did precisely that from the textual synopses of the hits, and I wasn’t going to take a subscription just on the off-chance.

So what gives here? If product managers are adding support for exact phrases then why are they not working? Why isn’t their absence being picked up during testing? Are the testers listening to software designers rather than reading functional specifications? A possible explanation is that the guts of the search engine is being brought in from some other company, and the testers test only the software developed in-house (well, ignoring the aforementioned help link) rather than testing that the whole system functions correctly as far as an end-user is concerned.

Why am I complaining? Well, the odd thing in The Gazette case is that none of the textual synopses correctly showed the context of the hit it was delivering to me. The following is just the first four of those 17 hits, so see for yourself:

Publication Date     18 June 1858
The London Gazette, Issue 22154, Page 2989
woo3, of Dewsbury, in the said county, Share Broker, and John Worsnop, of Oleckheaton aforesaid. Painter, all the real and personal estate and effects, whatsoever and wheresoever, of them, the said Wi…
View The London Gazette, Issue 22154, Page 2989 page

Publication Date     3 April 1849
The London Gazette, Issue 20964, Page 1093
awarded and issued forth against William Wood, of Wad- dington, in the county of Lincoln, Licensed Victualler, will sit on the 27th day of April instant, at eleven of the clock in the forenoon, at the…
View The London Gazette, Issue 20964, Page 1093 page

Publication Date     9 November 1849
The London Gazette, Issue 21036, Page 3371
at the Birmingham District :Cour.t of Bankruptcy, at Nottingham, in order .to Audit ,the Accounts of .the Assignees of the estate and effects qf the said""bankrupt under the said Fiat, "pursuant to th…
View The London Gazette, Issue 21036, Page 3371 page

Publication Date     17 November 1848
The London Gazette, Issue 20916, Page 4152
NATHANIEL ELLISON, Esq. Her Majesty's Com- missitfrter of the Newcastle-upon-Tyne District Court of Bankruptcy, the Commissioner authorized to act under a Fiat in of Bankruptcy, bearing date the 25th


There isn’t even a single instance of “Kirk” visible, and so the only option is to load each page image and perform a local search via Ctrl+F. This is just not good enough!




[1] Titian (1490–1576), “Sisyphus”, painting (between 1548 and 1549); attribution: [Public domain], via Wikimedia Commons.