Blog RSS Feed

Track in Real Time What People Are Giving Up for Lent in 2017

February 27th, 2017

See the top 100 things people are giving up for Lent in 2017 on Twitter, continually updated until March 4, 2017. You can also use the Historical Lent Tracker to see trends since 2009, though 2017 is still in flux, so I wouldn’t draw any conclusions about 2017 yet.

As I write this post, with about 1,600 tweets analyzed, perennial favorites “social networking,” “alcohol,” and “chocolate” lead the list. My main question, given the current U.S. political climate, is how high Donald Trump will rank: he’s currently vying with smoking and sugar for #24–last year, as a presidential candidate, he finished at #82.

Look for the usual post-mortem on March 4, 2017.

Jesus is not your password

January 6th, 2017

At Christianity Today I have a piece today about bad passwords that Christians use: Beware of Making Jesus Your Password. I’m pretty excited that they kept the line about soccer.

Here I want to share the data behind the piece. The 32 million passwords come from the 2009 RockYou breach, available here. I used rockyou-withcount.txt.tar.gz.

The main list of passwords comes from (1) taking this list, (2) removing non-alphanumeric and leading and trailing numbers, (3) lower-casing the result, and (4) combining the totals. In the raw list, “jesus” is the 103rd most-common password; by normalizing it with these steps, it jumps to #30. The purpose here is to find the core part of the password. It’s good from a security perspective that people add leading and trailing (mostly trailing) numbers to their passwords, but they’re not so relevant here.

The list of “Christian” passwords is based on a different breach of a faith-based website. I pulled a bunch of patterns from passwords that were popular there.

Here’s the data behind the piece:

  1. normalized-passwords.zip. A list of 238,000 passwords following the normalization scheme I describe above. Every normalized password from RockYou that appeared at least ten times is here. Note that there’s extensive swearing.
  2. christian-passwords.txt. All 505 Christian-themed passwords.
  3. verse-passwords.txt. All 295 plausible verse references. Not all of them are actually references: for example, “daniel14” could refer to Daniel 1:4 (or even Daniel 14), but it’s most likely just someone’s name with the number “14” after it. So I don’t include it in the top-25 list that appears at CT.

These are my favorite tweets about it:

The four kinds of one-year Bible reading plans

December 29th, 2016

Bible.com currently has 91 yearlong reading plans. By graphing these plans with time on the x-axis (January on the left, December on the right) and verse on the y-axis (Genesis at the bottom, Revelation at the top), we can divide them into four categories, labeled below:

An 8x12 grid of 91 reading plans.

First are continuous plans, which proceed in a sequential pattern through the Bible. They’re at the top of the above chart, characterized visually by strong lines proceeding from the lower left (January / Genesis) to the upper right (December / Revelation). They tend to cover substantial portions of the Bible, usually either the whole Bible or the complete New Testament. They generally ask you to read four or fewer distinct passages per day, with each passage about a chapter long.

Next are semi-continuous plans, which are the hardest kind to define. They feature stretches of reading that immediately follow from what you’ve just read, coupled with days of reading that are… not what you just read. You can see some lines go up to the left, similar to the continuous plans, and then they stop or dissolve into smaller sections. Defining a continuous vs. a semi-continuous plan is a matter of opinion; you could argue that some of the plans I’ve classified here as semi-continuous should be continuous. The most-interesting one to me is #34, the Book of Common Prayer Daily Office, whose pattern follows the liturgical year and doesn’t resemble any other plan on the chart. (Ten years ago I plotted both years of the Book of Common Prayer on a similar chart.)

Third are chronological plans, identifiable by their unique pattern of parallel passages in Kings and Chronicles (the parallel orange lines). Not all plans in this category are strictly chronological, but they have a similar organizing principle that produces the recognizable pattern of parallel paths.

Last are selections, which follow no discernible pattern and typically involve shorter readings–a few verses from multiple places in the Bible rather than reading full chapters. This kind of plan often includes devotional content related to the selected verses. The average plan in this category covers about 1,800 verses per year, or about 6% of the Bible’s 31,000 verses.

All the plans together

Here’s what all 91 plans look like overlaid on each other:

Paths radiate from Genesis in January to Revelation in December.

Paths radiate from Genesis in January through Job later in the year, varying mostly in the pace you’re reading the bulk of the Old Testament. In other words, if your reading plan starts by reading Genesis at the beginning of the year, you’re likely to proceed sequentially through the Old Testament until you reach Psalms. Similarly, paths converge on Malachi and Revelation in December.

Where are the digital-first plans?

The similarity among these plans suggests to me that they originate from a print-first mindset: what’s convenient to read in a physical Bible. You’re not, for example, going to flip to passages all over your print Bible every day unless you have unusually strong dedication to your reading plan.

But digital reading plans don’t have this limitation. Thanks to transclusion, a digital plan can collect content from all over the Bible (and elsewhere) and show it to you in one place, which means you can read much more widely on a single day if you like.

For example: let’s read through the Gospels chronologically in one year–that’s about ten verses a day–and also include ten cross-references for each verse every day. During a year, you read 14,104 different verses, or just under half the Bible. Here’s what the plan looks like schematically; you can see the paths through the Gospels, and overall it doesn’t look like any other plan:

The chronological Gospels plan shows a different pattern from other reading plans.

And here’s what the content for a single verse from one day’s reading could look like (because there’s so much related text for each verse, this plan requires a little information design): the current Gospel verse is at the top. Below it are three columns: one for Old Testament references, one for other Gospel references, and one for non-Gospel New Testament references. I also included the same color-coding as in the other charts in this post.

Matthew 5:5 with three columns of text.

Beyond transclusion, interactivity is a key property of digital systems. You could add interactivity by allowing the plan follower (with an interactive plan, they’re not just a “reader”) to configure the number of cross-references they read. Or tapping a cross-reference could open other verses they’ve read or will read as part of the plan–letting them explore more deeply and follow patterns as they desire. Adding interactivity to a system opens new avenues for insight.

Reading plans for augmented reality or audio-only interfaces will similarly vary in form from existing print-based plans. I would suggest that they will probably be even more conversational and dynamic than what I describe here.

I don’t know what shape future digital Bible reading plans will take, and the current approach of converting print-first plans to digital will certainly persist. But new digital-first reading plans will also come into being, and it’s important to experiment with what new technologies allow.

Super Mario Run is an exploration of free will vs. predestination

December 27th, 2016

Super Mario Run, an iOS game released December 15 and downloaded over 50 million times in its first five days, is a game about free will. Its three play modes reflect three different approaches to the topic.

Kingdom Builder: Free Will

The first play mode in Super Mario Run is Kingdom Builder, in which you can buy (with coins you collect elsewhere in the game) over 100 buildings and decorations to place anywhere you like on a map–think Animal Crossing or SimCity. A few buildings unlock new characters and courses, but little of what you do here affects the rest of the game.

In effect, you have consequence-less, complete free will within the constraints of the building system: you can do whatever you want.

This kingdom has lots of Toads but few buildings.

Toad Rally: “Predeathtination”

In the second mode, Toad Rally, Mario runs on what looks like a standard course with enemies and obstacles. As in World Tour (the third play mode), he runs on his own; you only control jumping. The twist is that you’re racing against another player–with success measured not in speed but in coins and skillful jumping.

But you’re not competing live against another person; rather, you’re competing against the Mario Kart-style “ghost” of another player who’s previously completed it (aka, “asynchronous multiplayer”). However, you’re not even really competing against them–you’re competing against an artificial intelligence (AI) that plays like them based on their historical playthrough. If you play the same course against the same player repeatedly, their performance and score will vary slightly.

One thing doesn’t change with every runthrough: the deaths of the other player. And that’s where predestination (in the non-salvific sense) comes in.

Luigi waits for Mario to appear on the screen.

In the above image, the current player as Luigi (in green, lower left) has run ahead of Mario and is waiting for him to appear on the screen. The circle with Mario in the upper left indicates that Mario is off the screen to the left and hasn’t arrived yet. Now look at the small “X” bubble near the yellow square on the right of the screen–it indicates that Mario will soon die there. (You can see a similar X bubble at the bottom of the screen, where Mario previously died.)

Mario jumps to his death in the flames.

This image shows the scene three seconds later: Mario has met his fate, leaping to his fiery death (the bubble he’s in will transport him back to safety). Luigi can do nothing about it–no matter how many times you repeat the course, Mario will die at this spot, his algorithmic doom determined when you first decided to challenge this player.

While your character (here, Luigi) has the jumping freedom and the constant risk of death featured in World Tour, the AI competitor doesn’t–they’re always going to die at the ordained time and place.

World Tour: Partial Free Will

The third play mode, World Tour, most closely follows the traditional Mario experience of running and jumping. However, unlike a traditional Mario game, in Super Mario Run, Mario constantly runs to the right with no input from you–you can control when he jumps by tapping the screen, and that’s about it. (I’m glossing over some subtleties.) Thus, you have limited options to explore a course when Mario is running relentlessly to the right, especially compared to the near-limitless freedom that Mario usually has in his other games.

Mario runs automatically.

In part, this control constraint reflects iOS’s touchscreen interface. Nintendo’s Shigeru Miyamoto says, “Our feeling was that by having this iPhone game in which Mario runs automatically and all you do is jump, we would actually have quite a bit of leeway to be able to create something that’s simple and easy for anyone to play, but still has some of that challenge and skill that super players desire…. By taking that approach, it would give even beginner players an opportunity to get a taste for what’s fun about the more skilled style of Mario play.”

From a game-design perspective, the single-tap interface justifies the automatic running and meets Nintendo’s goal of satisfying both novice and expert players.

However, from a philosophical perspective, World Tour reflects a hybrid of the complete free will of Kingdom Builder and the determinism of Toad Rally. Mario runs without input from you, but Mario needs you to jump, reflecting a more-constrained system than the usual Mario game has. It’s not quite the total freedom of Kingdom Builder or a typical Mario game, but neither does it feature the deadly determinism of Toad Rally. It presents a middle ground between the other two play modes.

Conclusion

The Wired article mentioned above says that a different team developed each play mode, and I suggest that each team, by accident or design, produced a distinct exploration of free will in each mode.

As you may have guessed by now, this post is mostly an attempt to justify spending $9.99 on Super Mario Run. Also, “predeathtination” (a word, oddly, that Google has never seen).

Visualizing Bible chapter similarity with Quid

December 22nd, 2016

Quid, a natural-language processing and visualization startup*, last month produced a network graph of chapter similarity in the King James Bible. It does a good job clustering the gospels and epistles in the New Testament, though I might argue that you can largely distinguish them simply by the presence of “Jesus +said” vs. “Jesus -said”. Their full post details their methodology and colors the same visualization several different ways, including by sentiment and by popularity.

Read the article on Quid's site.
Credit: Quid

* They actually call themselves “a platform that searches, analyzes and visualizes the world’s collective intelligence to help answer strategic questions.”

How violence, disease, and Donald Trump are signs of the end times

December 9th, 2016

At least according to searches for end times on Bible Gateway:

Bible Gateway searches for end times peak during certain events.
(The farther away the point is from the center of the circle, the more searches there were on that day.)

This image comes from The Three Kinds of Popular Keyword Searches on Bible Gateway, which talks about the three keyword search patterns Bible Gateway sees: perennial (not much variation throughout the year), event-driven (spikes around news cycles, like end times), and holiday-themed (spikes at the same time every year).

Bible Gateway doesn’t see much of a fourth pattern I might expect: a “seasonal” pattern that’s similar to “holiday-themed” but is more spread out and not driven by a particular day. For example, plotting Google Trends data for jacket using a similar methodology as above yields a much longer-lived rise and fall, where 75% of yearly searches happen between September and March, but there’s not a single, strong peak:

Google searches for jacket peak in the fall and winter.

I think–but haven’t been able to locate independent research to prove–that most search keyword data will fall into one of these four patterns (perennial, event-driven, holiday-themed, and seasonal) when plotted on this kind of chart.

The immediate inspiration for this visualization was Google’s Rhythm of Food (also see the associated Wired article), which explores this kind of seasonality in food searches on Google: they identify “seasonal” and “holiday-themed” patterns. Rhythm of Food plots absolute numbers of queries, so you can see that searches for kale have been falling since 2014 (it’s clearer in the animation if you click through):

Google searches for kale peaked in January 2014.

My interest in the Bible Gateway data wasn’t to track absolute numbers of queries but to explore yearly continuity, so each point on the Bible Gateway chart reflects the percent of queries on that day for the year (smoothed over the preceding and following three days). The downside of using percentages is that they can magnify small absolute differences.

Bible Gateway’s 2016 Year in Review lets you explore these patterns for 1,000 popular queries on Bible Gateway in 2016.

The Washington Post spins this report as More people searched the Bible for “end times” because of same-sex marriage than because of Trump.

If you’re curious how these thousand searches fit the concept / quote / entity / story framework I proposed for Bible queries in 2011, here’s how I categorize them:

Type Description Example Count
concept “Show me verses about…” love 593
quote “I’m looking for a specific verse.” I can do all things 253
entity Person or place Abraham 101
story Shorthand for a longer passage David and Goliath 53

“Data and the Bible Online” Interviews

November 21st, 2016

BigBible has been running a series of interviews called “Data and the Bible Online.” The latest is with me.

A recent interview in the series is with John Dyer. In part of his interview, he says, “I do worry that even for those of who value Scripture as part of their spiritual lives, it’s easy to confuse access to the Bible (i.e., having an app installed) with wisdom, maturity, or formation.” Just today I read This $1,500 Toaster Oven Is Everything That’s Wrong With Silicon Valley Design, with the tagline, “Automated yet distracting. Boastful yet mediocre. Confident yet wrong.” That tagline could well describe the future of digital Bibles.

In the article, Mark Wilson writes:

June [the oven] is taking something important away from the cooking process: the home cook’s ability to observe and learn. The sizzle of a steak on a pan will tell you if it’s hot enough. The smell will tell you when it starts to brown. These are soft skills that we gain through practice over time. June eliminates this self-education. Instead of teaching ourselves to cook, we’re teaching a machine to cook. And while that might make a product more valuable in the long term for a greater number of users, it’s inherently less valuable to us as individuals, if for no other reason than that even in the best-case scenarios of machine learning, we all have individual tastes. And what averages out across millions of people may end up tasting pretty . . . average.

Bible software can fall into these same traps, especially if an AI is involved. As Nicholas Carr likes to remind us, the more you automate something, the worse you become at it. Or, in Idiocracy terms:

Smartspeek: We think for you. We speek for you.
Image: 20th Century Fox

On the other hand, I can’t say that the American church succeeded in developing believers’ “wisdom, maturity, and formation” even before digital technology increased the Bible’s availability and immediacy. Matthew Block writes in First Things that evangelicals in particular misapply the Reformation idea of Sola Scriptura:

many Christians seem to think saying Sola Scriptura is the ultimate authority somehow means it is my personal “solo” reading of Scripture that is authoritative. They reject the witness of the Church down through the ages in favor of a personal, private understanding of Scripture (which is not at all what the reformers meant by the term “Scripture alone”).

Digital Bibles will accelerate this process by emphasizing personal application and understanding of Scripture, possibly–but not necessarily–building a theological echo chamber in which an AI can present you internally consistent interpretations that nevertheless fall outside what most would consider the bounds of orthodoxy.

Dyer argues elsewhere that embodied practices could serve as one antidote to these virtualization trends seeping into culture and consequently the church. We already see a secular reaction against the always-connected mentality with organizations like Time Well Spent, and Dyer suggests that the church has the tools it needs to speak to people both discomfited and empowered by technology.

As the U.S. moves to a post-industrial Christianity likely characterized by increased fragmentation and polarization, and as gatekeepers shift–Christianity Today recently chronicled the rise of parachurch women’s ministries that counterpoint the decline of traditional Protestant denominations–developing (or rediscovering) a solid theology around embodiment and presence will become increasingly important to the church. I lament in my BigBible interview the paucity of formal theological education related to digital engagement: the need to equip pastors and others to understand and respond deeply to people’s relationship with a technology-mediated existence. I believe the American church will need to grapple with this relationship in the near future–if only because the wider culture will also struggle with it.

Topics for the U.S. presidential election

November 18th, 2016

The below chart lists the most-popular topics on the topical Bible on the days after the U.S. presidential elections in 2012 and 2016.

End times, praying for government, and honoring leaders top this year's list of topics.
* “End times” includes the following topics: signs of the end times, end of the world, end of days, the end of times, in the last days, last days, the apocalypse, judgment day, america in the end times. “Honoring leaders” includes: honoring leaders, submission to human authority, obeying authority, respecting authorities, respecting those in leadership. “World government” includes: world government, new world order.

Topping the list this year is “end times,” which includes several related topics. The “end times”-related “the antichrist” and the 1990s-vintage “world government” also jumped a number of places. The Google Trends report for end times shows spikes around the 2012 and 2016 elections, though world events (especially in the Middle East) regularly increase “end times”-related searches.

Power of words” also jumped up a few places; during the October 7 news cycle around the Donald Trump/Billy Bush video, “power of words” was the fourth-most-popular topic, behind voting, marriage, and forgiveness.

Christianity Today puts this data in more context in their article The No. 1 Bible Verse and Top 25 Topics of Trump’s Election.

Emoji in the news

July 13th, 2016

I have two emoji-related pieces out today (since World Emoji Day is coming Sunday–mark your calendars, or, alternately just use the calendar emoji (📅), which on iOS says July 17).

The first is Emojis Reveal How Minorities Tweet the Bible Differently at Christianity Today, which looks at how Bible verses tweeted with darker-skin-toned emoji differ from those with lighter skin tones. I felt the topic differences between the two groups were especially stark (“money management” vs. “long-distance relationships,” for example).

I couldn’t find a good place for the following chart, which shows overall which emoji are most associated with darker and lighter tones (as tweeted with Bible verses). Fingers pointing upward (👍, 👆, ☝) are apparently a light-toned gesture, while bowing deeply (🙇) is a dark-toned one. It makes me wonder whether bowing and upward pointing carry similar semantics for different groups. It’s also a little funny to me how fingers pointing every other direction except up (👇, 👉, 👈) are more common among tweets with darker tones. And I don’t know what to make of the dark-toned raised fist (✊) vs. the light-toned fist bump (👊).

Nail Polish, Bowing, and Writing are most associated with darker tones, while Santa Claus, Thumbs Up, and Finger Pointing Up are most associated with darker tones.

The data behind the article is available for you to work with if you’re interested. You may want to turn off the filter in the “Tone users” column if you want the complete list of verses and not just the more-common ones.

The second post, All the 🌲🌳 of the field will 👏, is at the Bible Gateway Blog and talks about how you can now conduct emoji searches on Bible Gateway. The search engine preprocesses your query, replacing certain emoji with English translations. As with all translations, a certain amount of semantic leakage is inevitable, so searching for 💀 uses the literal translation, “skull,” rather than how people often mean it: “death.” I was tempted to translate 🕴 (man in business suit levitating) as something related to the rapture, but I restrained myself. Instead, it returns no results.

The Unrelenting Positivity of Contemporary Christian Music

June 4th, 2016

Leah Libresco at FiveThirtyEight recently analyzed the lyrics of Contemporary Christian Music and found that they’re overwhelmingly positive, especially compared with historical shape-note (Sacred Harp) music:

Pairs of words, like life/death and sin/grace skew to the positive in modern Christian pop.
Source: fivethirtyeight.com.

In the article, Libresco talks to Peter Beck, who offers the following axes of Christian experience to explain how upbeat Christian pop songs (which in a church context often repurpose themselves as worship songs) neglect “Winter Christians,” or Christians who engage (“commune”) with God and the church but who see complaining or lamenting to God as part of their spiritual experience. Instead, Contemporary Christian Music appeals to “Summer Christians,” who can have a tense relationship with Winter Christians.

Two axes, high/low communion and high/low complaint, illustrate four kinds of Christians.
Source: Experimental Theology.

Winter Christians follow a rich history, with lament Psalms making up almost half the book of Psalms. Indeed, my analysis of online Bible notes from last year found that complaints and requests for help accounted for 43% of all prayers recorded in these notes. Beck’s model is static, but he adds, “Personally, I think both situational and dispositional issues are in play. Many Summer Christians have ‘dark nights of the soul.’ But I also think there are some people who are tempermentally Winter in orientation.” In other words, while you might have Summer or Winter tendencies, circumstances can push you to the opposite side for a time.

Types of prayers in 52 online Bible notes (bold indicates a complaint)

Type Percent Count
Examination (examining / devoting self) 35% 18
Intercession (help for others) 23% 12
Adoration (praise) 13% 7
Tears (sadness) 10% 5
Radical (boldness) 10% 5
Petition (help for self) 8% 4
Suffering (agonizing with others) 2% 1

Libresco also quotes David W. Stowe, who suggests that “when that secular pop music moved on from this fear [of nuclear war and annihilation], so did the Christian music.” I would further suggest that the “all pop is now Scandinavian pop” effect is no small part of this shift. John Seabrook describes circa-2015 pop music this way: “ABBA’s pop chords and textures, Denniz PoP’s song structure and dynamics, ’80s arena rock’s big choruses, and early ’90s American R&B grooves.” The hooks, or catchy melodies, for these songs almost demand upbeat lyrics to match their upbeat feel.

The extensive crossover between secular and Christian music is visualized at musicmap.info, which argues that Gospel music as a distinct genre (below, at the top in blue) effectively ends as a major force in 1968 with the advent of modern Contemporary Christian Music. Instead, each secular genre develops a “Christian” subgenre: Christian pop, Christian R&B, Christian metal, etc. Libresco doesn’t talk about differences between modern secular genres and their Christian equivalents, but such an investigation would be interesting: does Christian rap, for example, “complain” less than secular rap? Do certain genres speak better to Winter Christians than Christian pop?

Musicmap.info's genres include gospel, arranged vertically by time.
Source: musicmap.info.

For another look at the relative popularity of Christian music genres and artists over time, see Google Research’s Music Timeline.