The Best Presentation at TOC

My favorite presentation of the conference was from Bob Pritchett of Logos Bible Software, in a session titled, Network Effects Support Premium Pricing. I remember attending his presentation four years ago at the first TOC in San Jose, so I knew I didn’t want to miss his presentation this time. They’re doing amazing stuff at Logos. They face an interesting challenge, one that every author and publisher faces: How do you compete against free? In their case, they sell about 10,000 bible study ebooks. How much has the bible changed over the last two hundred years? Not much. But what Logos excels at is making this information more accessible than ever before. They take a database-centric view of their vast and ever-growing library of content.

When you purchase a book from them, you’re not just getting a static ebook, you’re buying into a dynamic, integrated online application environment that becomes richer with each new publication, and with each new member to their community. Even if Bible study isn’t your thing, check them out for future-of-publishing inspiration. I can’t do them justice here.

High praise indeed from somebody who isn’t necessarily into Bible study, but recognizes that what Logos is doing is really quite unique in the entire publishing industry. Our “database-centric views” are only getting stronger, so you can expect to hear more about this in the months to come.

(Astute readers will note some similarities between this and my upcoming BibleTech talk. But the audiences are quite different, so the content will be too. This talk will provide “a practical case study on semantically organizing reference material to support search and navigation, using a controlled vocabulary.”)

Abstract

Encyclopedias and other subject-oriented reference books frequently present the same content using different names: and users often look for this information using other names altogether.

The Logos Controlled Vocabulary (LCV) organizes parallel but distinct content in the domain of Biblical studies to integrate reference information and support search, discovery, and knowledge management. The LCV captures

• preferred and alternate terminology
• inter-term relationships
• term hierarchy
• linkage to other semantic information

The initial version of the LCV (now shipping in the Logos digital library platform) comprises some 11,000 terms, and continues to grow as more reference works are added. It also provides the backbone of http://topics.logos.com, a website for user contributions to terminology and content.

This talk will describe the building of the LCV, how we’re using it now, and how we plan to use and extend it in the future.

Keywords: controlled vocabulary, publishing, reference, terminology, SKOS

Some time back, Tim O’Reilly (The Architecture of Participation) echoed and applied some observations from Dan Bricklin (the Cornucopia of the Commons) about the architecture of Napster and  other significant web-based systems. The individual details are well worth reading, but here’s the summary form. There are several common models for how to build large datasets that are valuable to people:

1. Pay people to build it (Bricklin calls this “Organized Manual”). Examples include the original Yahoo! directory of the web, and the Encyclopedia Britannica. There’s an variant that represents smart algorithms rather than just human effort (Bricklin: “Organized Mechanical”): this is how Google has built its indexes. But it still represents a significant monetary investment by somebody who probably expects something in return.
2. Get volunteers (Bricklin’s “Volunteer Manual”): Wikipedia is the preeminent example here, along with Linux, the Open Directory Project, and a great many open source projects. People do this work because they value the end result, and the project coordinates and magnifies those efforts.
3. Architect in such a way that individual self-interest creates collective value.

Napster (the original peer-to-peer version) was proposed by Bricklin as a prime example of the third model: simply by listening to your music (within the Napster ecosystem), the default settings meant you were also sharing that music with everybody else. Quoting Bricklin:

What we see here is that increasing the value of the database by adding more information is a natural by-product of using the tool for your own benefit. No altruistic sharing motives need be present, especially since sharing is the default.

This is Bricklin’s Cornucopia of the Commons (an allusion to Garrett Hardin’s Tragedy of the Commons): a system designed in such a way that use brings overflowing abundance.

(You might think blogging and twittering are like this, but they’re not. Nobody tweets because it has direct, inherent value to them: instead, it’s an outgrowth of a narcissistic, self-centered open, generous belief that what i say might have value to others. Few of us would do it if nobody else was listening. )

Models for Data Creation In Biblical Studies

All that (and Napster!) is now history, and i don’t want to get distracted by the peer-to-peer model that made Napster so powerful (Bricklin argues that’s not the reason it succeeded), or the legal issues that led to its demise. Instead, i want to reflect here on how these principles apply to Biblical studies and software.

With Logos 4, we’ve launched a major expansion of our Biblical Knowledge, by expanding Biblical People, adding Places and Things, and building around the large set of concepts we call the Logos Controlled Vocabulary. This was accomplished through the Organized Manual method: we paid a bunch of people (me included) to architect and populate this data, in a major development effort that stretched over several years. You could view the vast network of links that make Logos more than just a collection of texts as an extension of the same principle (through the resulting software program doesn’t look so much like a database). It represents literally hundreds of thousands of hours of effort in book markup and design, along with lots of “Organized Mechanical” algorithmic work.

There are also lots of examples of Volunteer Manual projects related to the Bible. The Sword Project is like Linux for Bible software. e-Sword has a smaller group of developers, but the same framework of a volunteer effort which is given away. Open Scriptures is building a platform and API for others to use in building Bible-based applications. Web 2.0 efforts like YouVersion let people tie their reflections directly to the Biblical text, and numerous projects have sprung from the Wikipedia mold like Theopedia. My own SemanticBible projects are much more limited, but in a similar spirit.

Logos has been active with the Volunteer Manual approach as well. The Logos Topics website combines our Organized Manual data and architecture of topics with user-contributed extensions of additional terminology, links within Logos, and even links to other websites. This lets us do some neat things like extending the desktop application content through user contributions on the web. Like Wikipedia, these are altruistic contributions from people who want to share their knowledge with others.

Sermons.logos.com works in a similar fashion: if you’re a pastor who writes down your sermon, and you’re willing to upload and share it, lots of others (both on the web and in Logos software) can benefit from what you’ve created. This is closer to the Cornucopia of the Commons model, but it’s still a voluntary and indirect process: my sermon doesn’t get shared as a natural by-product of my preparation activity.

The Cornucopia and Bible Study

The interesting question to me is how to achieve the third model, where my own use of a tool provides a direct benefit to others through a network, not because i’m behaving altruistically but simply because the system is architected to work that way. This is closely related to the whole Web2.0 meme (can it really have been five years already?!?) of “software that gets better the more it gets used.”

One thought: lots of web sites use RefTagger to provide a nice pop-up of Bible text for their readers, a benefit that enriches the experience of visitors to their site. Twitter users can similarly use ref.ly to shorten Bible references, which, like RefTagger links,  in turn resolve to references on Bible.Logos.com.   Could those links be converted into data indicating, for example, the relative popularity of different verses, and then displayed back to users?

Aggregating users’ operation of Logos software (in a suitably anonymized fashion, of course) could also provide data on the most popular resources, searches, and topics, which could then be turned around into recommendations (”Looking for a Bible dictionary article on ‘marriage’? Here are the ones our users have found most useful ….”).

But none of these seem to me to accomplish the full promise of the Cornucopia of the Commons. There has to be more here than simply harnessing popularity (though sites like Digg and del.icio.us have shown how useful that can be). I’m still trying to imagine what data sets could be created by people who are already committed to Bible study, as a normal outgrowth of what they do anyway. Any thoughts? Please share a comment.

The background: i’m using Logos 4, Syntax Search, with the Andersen-Forbes Phrase Marker Analysis (henceforth AF). If this is new territory for you, you’ll want to start by getting some background from Mike Heiser’s tutorial videos here: the syntax videos are currently down near the bottom.

So here are the steps:

• Open the Search panel, and select Syntax
• Set the search to All Passages in Hebrew Bible: Andersen-Forbes Phrase Marker Analysis
• Click on Query to create a New Syntax Search
• In the new panel, click on Add Search Terms here and select Segment. You’ll also want to check Show Details at the top of the panel.
• On the right side of the Panel, open up the Parts of Speech section, and in the Morph Specification type “@N?[NO]” (or select  the corresponding items from the pop-up: part of speech = Noun, Noun Family = numeral and ordinal). You can then click on Search to run this (i got 7128 results), and click through the results
• You might want to refine this search in several ways:
  • Limiting the scope to a particular book
  • If you have a deeper understanding of syntax, you may want to elaborate the search specification to only include particular contexts
  • Adding Semantics=Quantity removes two results, which was a little surprising, but i didn’t take the time to figure out which ones

One result of all this is a new appreciation for concreteness in such forums: reading for the 100th time “Logos 4 is so slow” just doesn’t help as much as a careful description of what the circumstances are (including the hardware environment), what operation is slow, what “slow” means (a few seconds? a few minutes? it just “feels” slow?), etc.

So, in the spirit of concreteness and positive contributions rather than complaining, and following up on some blog-chatter about the recent “shootout” at SBL, i’d like to tackle a nice “pop quiz” i recently found here about keeping your Bible software skills sharp. The original context was performing these tasks with BibleWorks. But i thought it might be useful to identify how to perform these same tasks in Logos 4, both for honing my own skills (i’m not just an employee, i’m a user), and for other users. Please note that i’m not trying to start (or fuel) any “my software can beat your software”-type competitions, or take pot shots at other products (i assume BibleWorks can do all these tasks just fine). I just liked the practical, objective orientation of this list as a learning exercise.

So without further ado, here’s the test (reproduced verbatim except as noted), the process i used to tackle each item, and the results i got.

1. Find all ESV occurrences in Paul of words beginning in, but not the word in.

   Process

   I opened the Search pane, selected Bible Search, and set the search scope to a custom range “Rom-Phil” that i called Paulines. I selected ESV as the Bible, and used the wildcard search expression in* ANDNOT in.

   Result

   This returned 94 results in 84 verses, in about a minute (wildcard searches tend to be slower).

2. How many Greek OT verses are there with the forms g:kurios AND g:theos? [note: i haven't figured out how to render Greek in my WordPress-backed blog, so i've represented these the way somebody without a Greek keyboard would enter them into Logos. Just to be clear, this is an issue with WordPress, not Logos.]

   Process

   I opened the Search pane, selected Bible Search, and set the search scope to Old Testament (Gen-Mal). I selected Septuagint (with Logos Morphology) as the Bible, and used the search expression g:kurios AND g:theos.

   Result

   This returned 1367 results in 592 verses, in less than a second.

3. How many Greek OT verses are there with the lemmas g:kurios AND g:theos?

   Process

   Same steps as above, but using Morph Search this time. As above, i used g:kurios AND g:theos in entry, which the search dialog then converted to the proper Greek forms as before. Then i edited them to have lemma: (not greek:) as the prefix, and @N as the morphological class. There’s a minor gotcha here: you might think you could use search syntax like lemma:kurios in Bible Search rather than Morph Search, but you’d be wrong (or at least you wouldn’t get the results you expect).

   Result

   This returned 3319 results in 1335 verses, taking about 3 seconds.You can see some inflected forms in the last verse included in the screen shot.

4. What’s the difference between these two searches? BGM .?????? and BGT .??????[again, i can't reproduce the Greek, but i'd enter it in Logos as g:iesous]

   Result

   Not being a BibleWorks user, i don’t actually know what this means. Maybe it’s searching two different texts? If somebody can translate this for me, i’ll see if i can determine an equivalent.

5. How many times does the word translated “creation” in Genesis 1:1 appear in the Pentateuch?

   Process

   I assumed “word” here meant lemma. I opened the NRSV, clicked the Interlinear button to display the interlinear text, and selected “created” (not “creation”, but i assume this was the intent) in the English text, which highlighted the corresponding Hebrew term. Then i right-clicked on that term, selected the Lemma tab in the right-click menu,  and selected Search This Resource. That searched the whole Bible (2 seconds), but the question said only the Pentateuch. So then i just changed the scope from All Passages to Pentateuch (i may have created this custom range before, i don’t recall).

   Result

   14 results in 11 verses.

6. g:agape is most common in what book of the Bible, when judged with regard to the book’s size?

   Process

   Here i used the Bible Word Study: the Lemma section provides a small sparkline-style graph with relative frequencies. Clicking on this opens up a larger pie chart that’s much easier to understand, but the counts here are absolute, not relative. However, the bar and column chart options let you select various display options, one of which is number of hits/number of verses in book (which i think is the intent of the question).

   Result

   Since i had a nice option to export the graph to PowerPoint, that made it easy to reproduce the results here. The winner is 2 John (i had assumed it was 1 John!).

7. How many times does g:agapao appear in the LXX vs. the GNT?

   Process

   Again, Morph Search using lemma:agapao (which magically turns into Greek), over two ranges: LXX, vs NA27, all passages in both cases.

   Result

   LXX returned 272 results in 256 verses. NA27 returned 143 results in 110 verses.

8. g:agape is in how many Louw and Nida semantic domains?

   Process

   Since i know Volume 2 of Louw-Nida has a Greek-English index that lists the domains for each term, the direct way is to just navigate there (using the table of contents) and count. That felt a little like cheating, but i don’t know of another approach that’s closer to the spirit of the exercise.You can of course search the resource for the term, but that produces lots of additional hits. There may be some deeper search magic that could be applied here.

   Result

   Two domains: 25.43 and 23.28.

9. How many ESV occurrences are there of words beginning inter or enter?

   Process

   This is pretty much like task #1, but with the range as All Passages, and the search expression inter* OR enter*.

   Result

   This was a slow search (76 seconds), which returned 450 results in 413 verses (assuming “enter” was to be included in the set).

10. How many GNT verses are there with g:uper alone vs. g:uper as a prefix?

    Process

    Like the search above, once you appreciate that wildcards word in Greek too. So these are both Morph Search operations, one with (the result of) lemma:uper, and one with greek:uper*.

    Result

    As a lemma, 150 results in 135 verses (2 seconds). As a prefix, 209 results in 184 verses. The screen shot is of the latter.

Some closing reflections on this exercise: overall, though it took a little time, i feel like i have a much better understanding now of how to perform tasks like these in Logos 4. The g: prefix syntax for entering Greek terms proved very handy (both for using the software and for blogging about it!), and performed just the way i’d want it to. There are some subtleties about Bible vs Morph Search on original language texts, but they make sense to me in retrospect. These counting tasks aren’t really the heart-and-soul of Biblical studies, but of course all the counts reflect a concordance that gives direct access to the verses in their context. So i’m encouraged that Logos 4 was able to address all these tasks quite easily, quite quickly, and with what i will assume are accurate results (until somebody lets me know evidence to the contrary )

Update (12/3): please see the comments below for discussion of a few things i got wrong.

• for #1, the range expression should be Romans-Phile (Philemon, not Philippians)
• there are some subtleties about selection order in Logos that matter to getting the Greek text right for #2 and #3, as Mark Hoffman explains in his comment

Also, credit where due to Charlie Gibson, BibleWorks trainer, as the source for some elements of the quiz (thanks for pointing that out, Mark Ward).

• Technology in the Bible provides an overview, including some metaphors and domain-level groupings (medicine, agriculture, military, etc.). I found his collection on God’s use of technology particularly interesting.
• Biblical References to Technology catalogs many verses with a brief comment on the technology involved.

If you have Logos 4, you can easily play along using the Biblical Things feature (brief tutorial video), which provides a comprehensive list of references for all the physical, depictable artifacts of technology (though not more abstract things like metal refining techniques).

This kind of broad study across the whole of Scripture can provide new perspectives on things that, in their immediate context, often go right by us.

• Biblical People I (overview) and Biblical People II (people diagrams)
• Biblical Places: overview, maps, dynamic maps
• Biblical Things

Logos 4 sports an entirely new interface to make it easier than ever to find what you’re looking for and keep your study space organized and effective. There’s a wealth of new, visually oriented resources, and better controls for working through the enormous space of resources Logos makes available. There’s even an iPhone app for no extra charge!

That’s the marketing view (and i stand behind it). But this means much more to me on a very personal level. It’s been almost 3 years since i came at Logos, and this will be the first time most of my work has seen the light of day. Specifically, Logos 4 contains the work of my colleagues and me in several new areas:

• Biblical People, which organizes information about the 3300 individuals, groups of people, and deities named in the Biblical text. It includes a comprehensive list of references, their family relationships, links to dictionary articles, and links to related items. It also includes family tree and story-based diagrams. And everything is hyperlinked.
• Biblical Places includes all the same kinds of information for 1200 named places from the Bible: cities, regions, even geographic features like rivers and mountains. Along with the data, there are 60 new high-resolution maps commissioned by Logos and covering the major Biblical events, as well as a mega-map that shows all the places together.
• Biblical Things describes the physical objects of the Bible: animals, plants, body parts, clothing, food and drink, and much more, as well as specific items like Noah’s ark and Goliath’s sword and weights and measures. There are more than 1000 objects here, which also bring together thousands of images from across the library.
• There’s also a new collection of high-resolution infographics illustrating different aspects of the Biblical world (and i’m extra proud that the bulk of this work was managed by my wife Donna)
• In additional to regular word search (which is much faster than ever), under the hood is the Libronix Controlled Vocabulary (LCV), working to organize 11,000 different subjects in the Biblical studies literature and coordinating information across the library.

So if you’ve been following my posts on the Bible Knowledgebase … well, now it’s here. I can’t overstate how important i think this is: this is quite literally the first time in the centuries-old history of Biblical studies that this information has been made available in this way. The LCV isn’t quite as visible (yet), but it’s also an important organizing feature that will continue to grow in power going forward.

I hope you’re catching my sense of excitement about these new resources (and this says nothing about all the hard work of my dozens of colleagues in other areas). I hoped i’ve piqued your interest to learn more about Logos 4. It really is a watershed event in Bible software.

Obligatory disclaimer: i work for Logos and highly value what i do there. So i’m not the least bit objective about this. (more detailed disclosures)

• ease of use
• continuous (ubiquitous) availability
• low price

For those of us working on what i’d claim is the Cadillac of Bible software (dating myself with that standard of reference!), namely Logos, this raises the obvious question: what’s “good enough” when it comes to Bible software?

One answer, of course, is how many people are currently voting with their wallets and (even more importantly) with their attention: that is, no software at all. Many Christians simply don’t read their Bibles, and of those that do, many use the old dead trees format, perhaps the shining example of “cheap and simple”.

But clearly technology has gone well beyond paper in offering accessibility to Bible readers: so where’s the sweet spot? Let’s organize the discussion around several levels of capability, with increasing sophistication:

• Free Bible texts available on the web (Bible Gateway, bible.logos.com, and many many others)
• Free software, typically including multiple Bible versions, search, perhaps some additional titles: e-Sword, the Sword project,
• “Entry level” packages for a modest fee: iLumina, PC Study Bible, QuickVerse, along with mobile software from vendors like Olive Tree and Laridian.
• Software for professional-grade study of original languages, commentaries, theology, etc.: here i’m thinking of Bible Works and Accordance.
• The library approach to Bible Study taken by Logos (my employer, so i’m not neutral on this point) offers packages at a range of levels and prices, so it both covers the spectrum above. But (to my mind) it’s in a class by itself because of the breadth and depth of its offerings.

(Just so you know, i took most of these from this survey of Bible software programs, rather than my own experience: i haven’t used all these packages.)

How do these systems stack up against Wired’s criteria for accessibility? Well, clearly free-on-the-web is the winner, as long as all you require is reading a Bible text. The cost is zero, it’s easy to find a passage (if you know the reference) with a search box, and it’s as available as the Internet. If you need genuine ubiquity (like on your smartphone) you can still get free software to meet that need, though with the added “cost” of installing and maintaining them. The entry level packages offer more functionality (other books, parallel translations, perhaps additional study helps), though they don’t really change the basic criteria. So you might wonder whether they’ll eventually get eclipsed by their “good enough” free counterparts.

The professional-grade software packages offer much more sophistication and many more resources than the free/modest cost packages, but those capabilities come with a significant price tag. Users with professional requirements really can’t get the same capabilities with a Good Enough alternative today, but such users represent only a portion of the larger market. The longer-term question is how the market for professional grade software will develop over time, given the continual pressure from free alternatives, and other general resources like Google Books and Amazon’s Kindle.

In some respects, Twitter seems like another replay of several earlier IT pyramid schemes: first bulletin boards, then websites, then blogs. The earlier adopters get the lion’s share of attention, and the people who follow them get the initial benefit of more information and (maybe) more access. But as the pyramid grows over time, the volume of communication becomes unmanageable. Then we evolve new schemes for managing the flood: bookmarks, feeds and feed readers, tags spaces (hash tags for Twitter), and other meta-schemes. At some point we ought to stop and ask whether the new communications services and the additional complexity and overhead they impose provide enough benefit to justify their cost.

I’m humble enough to acknowledge that many smart and productive people say they get a lot of benefit from Twitter, so maybe i really am missing something by remaining doggedly tweet-free. But the latest issue of eWeek (you may need to download their reader to access it) has a nice review of several enterprise-oriented social communication services (SocialCast, Socialtext, and Huddle) that’s making me rethink my position. These have several distinct differences from Twitter and Facebook that make them more interesting to me:

• Narrower scope of communication: While broad services like Facebook and Twitter provide access to a universe of information, that can become overwhelming, particularly when the universe is talking back at you. Limiting the conversation to what’s happening in our company has a lot of appeal.
• Better business information: getting the right mix between quantity and focus about what’s going on in your company has always been a difficult challenge in my business career. The people who have the most to say are also very busy, and of course they already know what they know: so there’s asymmetry, with a cost to them in producing information while most of the near-term benefits accrue to their listeners.At the same time, there’s no easy way to predict what information might be useful to whom, so generating lots of it makes sense provided there are effective ways to filter it.
• Distributed leadership: personnel at all levels have useful things to offer your company, if you can just break them out of the stovepipes of departmental structure and management hierarchy.

Here’s how i might see this playing out at Logos. Our business structure is largely typical for a software company of our size: there’s a sales and marketing division with some people who travel a lot and who (along with inside sales) provide a lot of our revenue, a customer service group that deals with users and their problems, the programmers who make the application work, a smaller group (where i work) that’s developing new features and data sets, a variety of support and infrastructure people who keep IT, finance, etc. running smoothly, and of course a management team that’s steering the corporate ship. All of these groups have front-line access to information that might help the rest of us do a better, given the right kinds of access.

One of the challenges is to keep the information we all broadcast a little more targeted. While some amount of personal interest makes all the hours we spend at work more fun, too much information about somebody’s disappointment that their favorite team lost, or how much they hate paperwork, the lousy sandwich they had for lunch, etc. would start to make this just like Facebook (and i don’t think having employees reading Facebook during work time is a good strategy for productivity). But i’d love to hear:

• from front-line sales people: a conference sales pitch that really hit home with people; the reason somebody just gave for purchasing our software (or not purchasing it); a complaint from a loyal customer
• from customer support: the common problems that get raised over and over, and that would make make for a much happier user experience if we fixed them upstream
• from support staff: the challenges that we can all help with
• from programmers and from R&D: exciting new discoveries, tips and tricks
• from management: new things we’re learning and thinking, big picture business issues, where we see our business heading

Many of these things get communicated now, just more formally, and therefore less frequently, and to a more selective audience within the enterprise. Moving toward Twitter-style microcontent, just a little more focussed, might provide the right mix to get me tweeting.