The home page of ChurchofJesusChrist.org, the generally very useful and professional website for The Church of Jesus Christ of Latter-day Saints, currently touts the improved search feature in the Gospel Library app, as shown in the image at the right. However, if you’re trying to use it for research in the scriptures, there are some serious gaps that need to be understood. I’ve made multiple efforts over the past few weeks to point these problems out to the IT team responsible for the app, but I think they have been swamped with the roll out, so I’ve only received standardized responses so far. Until the bugs are recognized and fixed, users need to be aware of what can go wrong. There are some major gaps that can lead you to miss the things you are searching for. Any scholarly work requiring accurate search results may be impaired.
For example, today I needed to find verses in the scriptures that refer to Michael, as in the archangel Michael. Searching on the app seems to work fine at first glance. Here are the results I get when I select “Scriptures” under “Collections” and then select “Doctrine & Covenants,” after scrolling down just past the first listed hit (Section 78) to show the lower part of the list:
See the results for Section 29? There are ZERO hits, and yet, the name Michael from Doctrine and Covenants 29:26 is proudly displayed, though when you click on that hit, it takes you to the top of Section 29 and does not show any matches. The app seems to think there are zero matches to display, but it obviously found one for that section.
In fact, most of the displayed hits appear to show one fewer match than actually occurs there. If you click on Section 128, for example, you will be taken to verse 21 and you will see the word “Michael” highlighted twice in that verse, with “1 of 2 matches” listed at the bottom of the screen. However, there are actually 3 matches, but the app is oblivious to the first one in verse 20. Strangely, sometimes the “missed” match is the first to occur on the page, and other times it’s not. And not every displayed hit is missing a match. Some hits with just one match will display just one match, not that troubling zero.
If everything were just off by one, the underlying problem might be a simple error where a redundant subtraction of 1 needed to be removed or something easy. But when results become erratic like this, it might be a deeper, painful problem in the database or core software, which may be why it hasn’t been fixed yet after my repeated efforts to point this out.
I experienced these problems on my iPhone 11 with the latest operating system (IOS 15.6.1) and the current app version (6.2.1 12757). I also tested an iPhone SE also with the latest operating system and app version, and it had the same problems. My old iPhone 6+ running with Gospel Library app version 5.12.6 (the latest available for this old hardware) does not have this problem and searches properly. Further, it gives results in a logical order: Section 27 first, then 29, 78, 88, 107, and 128. The order of the search results in the new app is puzzling and less useful.
A clumsy workaround is to go to the page of any hit of interest and then retype your search term and choose the “Find … on page” item that will appear below your search term. Then you can manually go through each match on that page to pick up any missing items. This, of course, is tedious.
“But wait,” you are thinking, “if the app is so buggy, I can just go to the scripture section of the Church’s website and search for terms there.” And yes, you can, but prepare yourself for a long and bumpy ride. While the app found 6 sections of Doctrine and Covenants that mentioned Michael, my search for the name “Michael” using a Firefox browser at ChurchofJesusChrist.org > Libraries > Scriptures > Doctrine & Covenants gave me 6 pages of results with 10 hits listed per page, a total of 53 hits, with nearly every hit being a section of the Doctrine & Covenants. Here are the first ten hits displayed:
So apparently the search engine for the website thinks “Michael” occurs in Sections 2, 4, 24, 31, 75, 87, 103, 116, and 131, none of which were hits for the same search in the app. None of these sections appear to mention Michael at all. I started working through the rest of the 53 listed hits, and soon found sections listed that were in German and Spanish, and later the Fante language (an African dialect related to Twi) and then Tok Pisin (New Guinea Pidgin, a commonly used pidgin on Papua New Guinea). A lovely international journey, but not a very efficient way to search. It wasn’t until I got to the 4th of 6 pages of search results that I found the first “hit’ that actually had a “Michael” in it, Section 128, and then Section 78 at the bottom of that list, plus more on the 5th page, which listed Section 29, a hit in English as well as in Tok Pisin, and also Sections 27 and 88. But none of these pages of hits listed Section 107. And strangely, when I went to the second page of the search results, the list of six pages at the bottom expanded to 10. If I clicked on any of the new pages of results (7, 8, 9, or 10), I was taken to the results of an image search that told me there were no results for “Michael.” But this didn’t always happen — sometimes page 3 of the search results would show the inflated list, and other times neither page 2 or page 3 would do that.
This mishmash of English and foreign language search results for an English search, the abundance of false hits, the strange order of the results, the instability of the results (sometimes showing 10 pages of hits instead of “the real” but still wrong 6) may all point to serious flaws that may be difficult to debug. But I’m grateful that Tok Pisin is being given more attention and enjoyed the opportunity to recognize a few easy cognates on a couple pages of the Doctrine and Covenants in that language.
I should also note that one of my first complaints to the app team about the new version was that searching no longer matches related word forms. For example, if I want to find words related to “comprehend,” I can’t just search for “comprehend” and find “comprehends,” “comprehendeth,” “comprehending,” and “comprehended.” Section 88 has 8 occurrences of the root “comprehend.” But when searching with the new app, it finds only two: “comprehend” in the heading and in verse 43. Searching still wasn’t great in the older versions: on my old iPhone 6+ with an old version of the app, when I search for “comprehend” in Doctrine and Covenants, I get 5 matches in Section 88, including “comprehend” and “comprehended,” but there are also three occurrences of “comprehendeth” that are missed. There should be robust stemming features that users can select to find related word forms for a common root, including KJV-like “eth” and “est” forms. But that should be an option one can turn on or off. Better search means better research and makes the app more useful.
Here’s hoping that the IT team working with the search functions in the app and on the website will soon address these bugs and bring us back to more reliable and stable search results. I hate to mention these problem since they have done so much great work in some many aspects of the app and the website, but for students of the scriptures, the ability to seek and successfully find is of core importance.
Update, Aug. 26, 2022: I just received an email from the Gospel Library app people acknowledging that there is a search bug:
There is a problem with the new search, where it does not find words that have footnotes. The developers are aware of it and are working on it. Please be patient as we try to improve the Gospel library App.
–Gospel Library Response Team
Ah, so if there’s a superscript letter next to a word indicating a footnote, the word won’t appear as a match when you go to the page that is shown as a hit. That explains some of the trouble.