THE KORR CONCEPT: My casual-songsheet reinvention has three main ideas:

1. Re-orient songsheet formatting around musical timing — beats and measures — rather than lines of prose.

1. Creatively convey vocal phrasing through spatial and phonetic representations of words, syllables, and sounds.

1. Use design elements to improve legibility and wayfinding.

For example, here’s the Korr Format songsheet for Creedence Clearwater Revival’s “Have You Ever Seen the Rain”:

WHY IT MATTERS: Most musicians and singers learn songs using what I call casual songsheets: internet chords, campfire printouts, Rise Up Singing — basically anything that’s not sheet music. But the dominant casual-songsheet formats fail musicians and singers in two key ways:

• They make it hard for musicians to know when to change chords.

• They make it hard for singers to know when and how to sing a given line or phrase.

In this post I’ll unveil my core innovation: beat-based songsheets. I’ll cover vocal phrasing and wayfinding innovation in follow-up posts.

Grab your guitar/uke, warm up your voice, fire up Spotify — and let’s nerd out on songsheets. 👩‍🎤🤘🎸🎼🎺🧠

Current Songsheet Formats

Before diving into my new approach, I’ll briefly recap the dominant casual songsheet formats.

Internet Chords

This is the most common songsheet format. You see the Internet Chords format1 on sites like Ultimate-Guitar.com and others stretching back to the pre-browser days.2

Here’s the the Internet Chords format for a different song, The Temptations’ “My Girl”:

The crucial thing to notice is that Internet Chords songsheets:

• Revolve around the lyrics rendered as discrete lines of coherent prose.

• Ignore the song’s musical timing.

That is, the songsheet creator tries to make each typed line of lyrics a complete sentence or phrase if you read the line as prose, regardless of how that line corresponds to the music’s timing. As a result, chord placement is haphazard, confusing, and often incorrectly/arbitrarily aligned with the lyrics.

Rise Up Singing

The Rise Up Singing format3 is used in the seminal sing-along songbook (first published in 1988) and its 2015 sequel, Rise Again4.

Here’s Rise Up Singing’s approach to “My Girl”:

While the chord placement is different than in Internet Chords, Rise Up Singing shares that format’s approach of rendering lyrics as discrete lines of coherent prose. (Rise Up does try to convey some musical timing, but it’s incomplete and unintuitive.)

Introducing Beat-Based Songsheet Formatting

My core innovation is to reframe casual songsheets around musical timing rather than around lyrics-as-lines-of-prose.

Here’s the Korr Format songsheet for “My Girl”:

Rendering and reframing around measures and beats conveys key information that’s missing from the other formats:

1. When singing doesn’t start “on the one”

2. When chord changes and vocal timing are unusual or extra-precise

3. When a song’s consistent timing briefly changes

When singing doesn’t start “on the one”

In many songs, chord changes occur on a measure’s first beat (aka “on the downbeat” aka “on the one”). But for any given lyrical line:

1. Singing does not start on the one — instead waiting a bit to create rhythmic, anticipatory, or propulsive effects.

2. Often, this causes the lyrical line to wrap to the next musical line. In this case, the words sung on a measure’s first beat are conceptually in the middle or end of — rather than the start of — the lyric/prose line. (This is the musical version of a poetry technique called “enjambment.”)

My Girl

We see/hear both techniques in “My Girl.” There’s a recurring pattern where David Ruffin waits to start singing lyrical lines until later in the musical line (“I’ve got …”, “When it’s …”). This wraps the lyrical line to the next musical line, so mid-sentence words — not the sentences’ first word — are sung on beat 1 (“… sunshine”, “… cold outside”).

This is all visible in the Korr Format:

In contrast, current songsheet formats always align the first word of a lyric line with the first beat of the music line, which leaves musicians and singers lost when the vocals and chord changes aren’t aligned:

With or Without You

U2’s “With or Without You” is another good example. Here’s the Korr Format version:

Zooming in on the verses, we see that Bono establishes a pattern of waiting several beats before starting to sing a lyrical line (“See the stone…”, “Sleight of hand…”, “Through the storm…”). These pauses help convey the song’s yearning feeling.

Again, the Korr Format makes that clear and easy to follow:

And again, the Internet Chords version is totally unhelpful:

The formatting incorrectly implies that each lyrical line’s first word is sung on Beat 1, and doesn’t convey that many chords are played on Beat 1.

When chord changes and vocal timing are unusual or extra-precise

Reframing songsheets around beats is also helpful for communicating chord changes and vocals that occur precisely on (or precisely between) specific beats.

Rolling in the Deep

For example, the Adele hit has unusual chord-change timing: In the verses, most chord changes fall on the 1-and and 3-and beats, which always feels later than you’d expect. I’ve heard this song dozens of times, and those changes still aren’t intuitive for me.

Beat-based formatting makes it much easier to follow this unusual timing:

When a song’s consistent timing briefly changes

Beat-based formatting also helps with brief sections that depart from a song’s previously established timing.

The Boxer

For example, Simon & Garfunkel’s “The Boxer” has a one-off section in Verse 4 that’s tricky to follow. In the second half of the verse, the melody, chords, and timing all change from the previously established patterns:

Beat-based formatting makes it easier to learn and follow these one-off changes.

An Iterative Design Challenge

The Korr Format is very much a work in progress. I’m constantly iterating — you can see different variations across the above screenshots — and don’t feel like I have a polished v1 yet. I also need a lot more feedback. (Which is one reason I’m writing this bloggie!)

Here are the biggest design challenges I’ve encountered.

Space Constraints

While the Korr Format could be used for any context, my focus is on printed songbooks for my sing-alongs. I use a two-column layout try to squeeze as much content as possible on each page (to keep the book length manageable), and try to fit each song within a max of two facing pages (so musicians don’t have to turn the page mid-song).

Both of these choices create space constraints. For example, I often combine two measures on one line so the song doesn’t spill onto a third page — but then the two-column layout forces me to squeeze words or beats closer together than I’d prefer.

Too Much Information

Between the chords, lyrics, and beats, there’s a lot of information to track. The other Korr Format elements that I didn’t cover in this post — vocal phrasing and wayfinding — add even more details to parse.

I hope that’s all intuitive enough to sub-consciously pick up in the moment, but I’m guessing I’ll need to pare back some ideas.

Inconsistent Beat Spacing

This is the biggest — perhaps unsolvable — challenge: as long as beats and words are shown together, the beats will never be consistent from line to line.

For musicians, it would be easiest to learn chord-change timing if beat spacing was consistent from line to line. For example, the “My Girl” chorus would be easiest to learn if the beat-based songsheet looked like this:

But adding words to the songsheet forces inconsistent beat spacing from line to line:

This happens because different words have different typographic lengths, even if they have identical musical-time lengths. For example, here “I” and “what” have the same musical length (one beat) but drastically different typographic lengths. So the beats’ positions have to shift from line to line depending on the lyrics’ typographic length.

What do you think?

I’d love feedback and ideas to keep improving the Korr Format. If you want to try it out, email dcsingalong (at) gmail (dot) com and I’ll send you a songsheet!

To make this process easier, I created a custom GPT: DictaBot, a digital assistant for recording and transcribing ideas.

Here’s a rundown of how to use DictaBot, and the problems it solves for me.

How to Use DictaBot

You’ll need a ChatGPT Plus subscription to use DictaBot (and other custom GPTs), and the ChatGPT mobile app.

In the ChatGPT mobile app, start a new DictaBot chat, then press the microphone button to start recording your voice.

When you’re done recording, press the Stop button. (I don’t know the max time limit; I’ve tested as long as 5 minutes with no problem.) DictaBot will transcribe the recording and populate the transcription in the chat’s Message field. Press the submit button to add the transcription to the chat itself. You can copy the transcription from there and paste it into other apps (or other ChatGPT chats) depending on what you want to do with it.

After adding the transcription to the chat, DictaBot should respond by saying only: “I hope the transcript was accurate. You can record another idea, or ask me to apply my GPT powers to the transcription.”

By default, ChatGPT would automatically summarize, edit, or otherwise riff on whatever you submit via the message field. But I’m using DictaBot for capturing raw ideas, not fleshing them out — so usually I just want the transcript, nothing more.

To prevent DictaBot from summarizing etc., I crafted the GPT’s configuration prompt so that it would reply only with the above response. I’ve found DictaBot returns only this response as expected about 90 percent of the time. (Anecdotally, hiccups seem to occur after a ChatGPT update.) Note that you can still type in messages/prompts if you want DictaBot to summarize or do other standard ChatGPT functionality.

Here’s the full DictaBot configuration prompt (aka “Instructions”):

This GPT’s main role is to act as a digital assistant for recording and transcribing ideas. It should be capable of quickly understanding and organizing spoken or written thoughts into a clear, coherent format. The GPT should prioritize accuracy and clarity in transcription. After a user submits their transcript, the GPT should respond with these words exactly (omitting the quotation marks): “I hope the transcript was accurate. You can record another idea, or ask me to apply my GPT powers to the transcription.” Do not respond to the transcript submission by repeating the transcript, or with any other words except for (again omitting the quotation marks): “I hope the transcript was accurate. You can record another idea, or ask me to apply my GPT powers to the transcription.” The GPT can respond in other ways if the user submits a text prompt that is not a recording transcription. In this case, the GPT should maintain a professional, supportive tone when responding.

Problems DictaBot Solves

The main problem DictaBot solves for me is enabling what I call Still Processing mode in the creation process, vs. Composing mode.

Still Processing is the stage/state when I’m working through rough ideas that I haven’t fully articulated in my head. In contrast, Composing mode is when I already have a clear sense of what I want to say, and I’m crafting structure and prose. (Shaping mode is roughly in between.)

In Still Processing mode, I find formats like handwriting and voice recording best for drawing out fuzzy thoughts with minimal friction or pressure. The problem with voice recording is you need a transcription for recordings to be useful.

Pre-LLM transcription solutions were pricey and/or a hassle, so I tried dictation features in Google Docs and Apple Notes, thinking they’d be a cheap record-and-transcribe solution. But I kept getting stuck and couldn’t make progress — because it turns out there’s a subtle distinction between dictation and recording.

Dictation’s format and functionality are really for Composing mode

• The format of a blank doc with a blinking cursor signals “write!”

• Dictation functionality can only transcribe a line or two at a time, not a long recording. So you really need to already know what you want to say; dictation can’t handle lengthy stream of consciousness riffing.

DictaBot solves this by keeping me solidly in Still Processing mode. I can thought-vom until I’ve captured the kernel of an idea, without fiddling with prose. As a bonus, I can do this on walks or in the car as well as at my desk. Later, the transcripts give me a starting point to shape and compose a draft.

Try It Out!

Give DictaBot a try and let me know what you think!

And yes, I used DALL-E to create the cute DictaBot logo:

I'm a sing-along impresario in my spare time, and I recently made a breakthrough discovery in hootenanny technology: The surprisingly awesome kazoo.

Kazoos have added a musical and participatory richness to sing-alongs that I never expected — and that never fails to inspire delighted grins in the audience.

In this post, I want to share my kazoo discovery: I’ll explain how I incorporate kazoos, why they’re surprisingly awesome, and how I discovered their potential. Hopefully you’ll be inspired to try kazoos at your sing-alongs — or to try it out at a D.C. Sing-Along hootenanny!

How I use kazoos in sing-alongs

I’ve incorporated kazoos into a bunch of songs, as a stand-in for a variety of instruments:

Horn Parts

The horn riffs in Ring of Fire and Sweet Caroline.

Synthesizer Parts

The synth riffs in Take On Me, Mr. Brightside (post-chorus), and Dancing on My Own (ditto).

Guitar Parts

The guitar riffs/licks in Girls Just Want to Have Fun, Brown Eyed Girl, and Respect; the solos in The Middle and Say It Ain’t So.

Whistle Parts

The whistle solo in Me and Julio Down by the Schoolyard.

Flute Parts

The panpipe riff in My Heart Will Go On and flute solo in California Dreamin'.

Steel Drum Parts

The (faux) steel drum riff in Under the Sea.

Why kazoos are surprisingly awesome

Kazoos are musically legit

When I first had the idea, I assumed kazoos would be a goofy gimmick: We’d use them for a song or two, but kind of roll our eyes and probably want to cover our ears.

But it turns out kazoos actually sound good.

Yes, they’re goofy. But they sound much more like an instrument than a toy, with a rough, distorted sound that’s more rawk than Raffi.

Kazoos turn non-musicians into musicians

I center my hootenannies on the singers (of all skill levels) rather than the musicians, but I think singers sometimes still feel apart from the musicians. Kazoos bridge the gap, letting non-musicians feel like they're playing an instrument.

Kazoos are super accessible: They’re easy to “play” (if you can hum, you can kazoo) and non-intimidating (they look like toys). Plus they’re just fun!

Kazoos expand and enrich the repertoire

Kazoos are the first viable way to add memorable/crucial riffs and solos — like the Take on Me synth riff — to the sing-along repertoire and experience.

Countless songs include these elements, which typically wouldn’t work in the sing-along context. Either nobody can play the part on acoustic guitar, or if someone can play the part it’s either too quiet or becomes too much of a performance. So you end up playing a song like Take on Me without its crucial riff — which subtly diminishes what makes it so beloved — or omitting the song entirely, as I did before the kazoo discovery.

But with kazoos, those riffs and leads are newly viable — and in a way that matches my sing-along ethos of being about the group, not the musicians.

One caveat: I’ve found riffs and short parts to be more successful than longer parts and solos. More people can more easily pick up the short parts: I can teach short riffs, and many are embedded in our collective memory. Whereas the Me and Julio whistle part is too long to teach, so it’s often just me and a couple audience members tooting that one, which is less participatory than I prefer.

Kazoos Keep Kids Engaged

Kids love kazoos.

They love making funny noises. They love having something to interact with. If they’re old enough to realize it, they love the feeling of playing an instrument. And for self-conscious tweens and teens who don’t want to be seen loving anything, kazoos help them tap into being a goofy kid again. (Sorry, blasé friend: Even if you’re tooting with a stone face, I can tell you’re into it.)

Of course this one is a mixed bag: Especially younger kids can be … let’s say overzealous in their kazooing, which can create cacophony.

The kazoo innovation process

I discovered the kazoo’s potential while thinking about how to introduce more fun participation elements to my hootenannies.

I remembered a cool makeshift instrument from my college a cappella group: Becky Bascom used a comb with a piece of paper in front of it to create a vocal distortion effect while singing the California Dreamin’ flute solo and Sweet Child O’ Mine guitar solos. I thought I might be able to use a kazoo for the same effect.

I ordered a batch of kazoos on Amazon to test myself. As soon as I did my first at-home test, I knew I had discovered something revolutionary.

I quickly validated the awesomeness with my music buddy Lawrence, then did a beta test at a House Hootenanny™. I brought kazoos to the next public sing-along, and the rest is history. I’ve since led 15-plus hootenannies with kazoos including the Takoma Park Folk Festival and Petworth PorchFest, and they’re a huge hit every time.

Tips for Sing-Along Leaders

Which kazoos should you buy?

I use these kazoos from Amazon ($12 for 20). The “reeds” are consistently fastened and loud. (I tried a 50-pack from the same manufacturer, but it turned out to be a different model and had a ton of duds. So I only by the 20-pack to be safe.)

These come in individually wrapped pairs; at the beginning of a sing-along I put a wrapped pair on every other seat, so participants know they’re fresh. I tell people to take the kazoos home and bring them to the next hootenanny they attend.

I don’t like the environmental implications of using cheap plastic kazoos, but don’t have a good near-term alternative.

How to play kazoo and guitar at the same time

I’ve tried both clamping my kazoo in a harmonica holder and taping a kazoo to my microphone:

I’ve settled on the mic-taping approach, as it’s easier to quickly switch between kazoo and vocals. But I’m never sure how well the taped kazoo gets picked up by the mic — it’s not an issue in small rooms, but definitely can be a challenge for larger spaces or outdoor sing-alongs.

I finally put on my product critique glasses and figured out why.

The Problem LinkedIn Should Solve for Job Seekers

In A concept model for prioritizing job applications, I shared a framework for prioritizing job applications based on our odds of getting a meaningful first look:

• Have a Referral-Willing Connection ➞ Odds: Plausible

• Have a Relevant Pool-Narrowing Factor ➞ Odds: Mid to Long Shot

• All Other Jobs and Scenarios ➞ Odds: Lotto Ticket

Given all of that, the main problem LinkedIn should be solving for job seekers is:

I want to easily find relevant jobs where I have a connection — including and especially 2nd-degree connections — because those are my best chance at getting in the door.

Alas, LinkedIn fails at solving this core problem for a number of reasons:

1. The concept of “relevant jobs where you have connections” barely exists.

2. Finding jobs via your connections pages requires tedious manual effort.

3. It’s inefficient to see if you have connections when viewing company pages.

4. Meantime, LinkedIn encourages you to waste time on low-value activities.

Grab some popcorn, and let’s dive in.

LinkedIn Frustration No. 1: “Jobs where you have connections” is barely a thing

My core frustration with LinkedIn is that it barely represents the concept of “jobs where you have connections” (again, the most valuable information for job seekers).

To illustrate this, follow me on a LinkedIn journey.

We start on the logged-in homepage — which lacks any jobs content, let alone high-value info about jobs where you have connections.

OK, let’s try the Jobs page.

Nothing useful at the top: It starts with “recommended” jobs, aka jobs based on your profile and search history. These don't surface connections or Pool-Narrowing Factors, so they are Lotto Tickets.

Next are a bunch more Lotto Ticket job lists: 

• “Jobs Where You’d Be a Top Applicant” (only visible if you pay $42/month for premium, but these are a trap — they're simply Lotto Tickets)

• “Job Collections” (themed Lotto Tickets lists)

• “Still hiring” jobs (who-knows list of Lotto Tickets)

• “Green Jobs” (industry-specific Lotto Tickets list)

Only then do we finally see LinkedIn’s sole list of jobs where you have connections.

But on closer inspection, that list isn’t really useful.

First, the list doesn’t appear to have any filter logic beyond "in your network" — nothing based on your profile, searches, saved jobs, etc. — so it surfaces mostly irrelevant roles and locations.

Second, its definition of "in your network" is too limited to be consistently useful.

It only shows 1st-degree connections who have a connection with "the hiring team" — which I assume means a LinkedIn Recruiter user who posted the job, shared the job, has the Hiring Manager role in Recruiter, or otherwise denoted themselves as on the hiring team.

This is too narrow. Rather, what you want to see is any 1st- or 2nd-degree connection at that company, full-stop:

• 2nd-degree connections — not 1st-degree — are often how you get a referral or introduction.

• Your connections may know relevant people at the company who are not denoted as the hiring team via LinkedIn Recruiter.

Frustration No. 1 Summary

To sum up, LinkedIn's core value for job seekers should be helping us find relevant jobs where we have connections. But it only has a single, buried view of this information — and the info it provides there is not very helpful.

LinkedIn Frustration No. 2: Want to mine your connections? Have fun manually clicking for hours!

Without useful job lists, job seekers’ next obvious path is to check if our 1st-degree connections have connections to relevant jobs.

Another alas: LinkedIn offers little help here, forcing us to do tedious manual busywork to mine our connections for jobs.

Let’s return to our LinkedIn journey, this time going to my wife’s connections page. Here I want to see surfaced any relevant jobs at her 1st-degree connections’ companies (i.e., at my 2nd-degree connections' companies).

Aaaand nope:

The connections page does not surface jobs.

Ok, so I need to manually investigate each 2nd-degree connection to see if their company has relevant jobs. Here’s what that process looks like:

• Naturally, the connections list page doesn’t display links to company pages. (See previous screenshot.)

• Instead I have to first go to the 2nd-degree connection's profile page (Click 1).

• Welp, that page doesn’t surface relevant jobs at their company, either 🤪.

• Oy, I guess I have to go to the company page. So in the person’s profile header, I click the logo of their current company to go to the company page (Click 2)…

… but, surprise! That doesn’t actually take me to the company profile page. Instead, it jumps me further down the person’s profile page, to their current Experience entry. When I click the company logo there, I’m finally taken to the company page (Click 3).

Then there’s one more click to view the Jobs tab (Click 4).

Frustration No. 2 summary

It takes 4 clicks and 2 page loads to see if a single 2nd-degree connection’s company has relevant jobs. Now multiply that manual effort by your closest connections’ thousands of connections 🤯.

It would be one thing if this were the free product. But I’m paying for LinkedIn Premium, and still get this frustrating customer experience.

What about that "Actively hiring" filter?

Eagle-eyed readers may have noticed an "Actively hiring" filter in the connections page screenshot above ...

Unfortunately, this is another mostly-useless paid feature:

• The list is narrowly defined as 2nd-degree connections who have denoted themselves as #hiring via the LinkedIn Recruiter service. (Again, what you'd want to see is all 2nd-degree connections whose companies have relevant jobs).

• Similar to the "Hiring in Your Network" list we saw in Frustration No. 1, these jobs aren't further filtered to your role or location preferences — so most of them are irrelevant roles and/or irrelevant locations.

LinkedIn Frustration No. 3: It’s inefficient to see if you have connections at a company

So far we've looked at jobs pages and connections list and profile pages. What about if you’re looking via a company profile page?

You’d hope/think LinkedIn would prominently surface if you have 1st- or 2nd-degree connections at the company you're looking at. Ready for this one?

The company profile header doesn’t show or summarize your connections at the company. Instead, it shows 1st-degree connections who follow the company page. I can’t think of a more useless piece of information.

At least in this case it's only one click to reach the People tab, which does show your connections at the company — though in a seemingly random order, so you often have to scroll to see them all and are never sure it's a comprehensive list.

LinkedIn Frustration No. 4: It encourages you to waste time

The first three LinkedIn frustrations are directly related to job seekers’ primary job-to-be-done (easily find relevant jobs where I have a 1st- or 2nd-degree connection). The final frustration is more general: LinkedIn encourages job seekers to waste time on low-value activities.

Job search theater

Let’s revisit one of the Jobs page screenshots from earlier, with some additional annotations:

This is all job search theater (a la security theater): information and activities that provide the feeling of job searching, but don’t actually help you find a job.

• List after list of Lotto Ticket jobs. The "Jobs Where You'd Be a Top Applicant" list is especially frustrating: As a prominent Premium feature (again, $42 monthly including tax), it makes you think you have a real chance at these jobs. After a while, you realize they’re just Lotto Tickets unless you have a connection or Pool-Narrowing Factor.

  • Update: After publishing this article, I learned I have a 1st-degree connection who is connected to the co-founder of Matter Neuroscience (shown in the "Top Applicant" screenshot above). (a) That would have been nice to know! (b) I'm guessing it's coincidence that this job showed up in my Top Applicant list, and that connections (inexplicably) aren’t factored into the Top Applicant logic.

• Data elements that imply relevance but are actually useless: School alumni at a company; review/response times.

• No way to filter or save preferences for automated job lists.

Digital distraction

Finally, let’s return to the logged-in homepage:

Have you worked hard in recent years to quit your social media / digital distraction habits? Well, get ready to relapse — because LinkedIn’s homepage is 100% time-wasting digital distraction:

• A classic, infinite-scroll social media feed.

• News headlines.

Job Seekers Aren’t the Primary Customer

As a job seeker, it’s frustrating that LinkedIn fails in these ways. But I understand it from a business strategy perspective: Job seekers aren’t LinkedIn’s primary customer — employers and recruiters are.

LinkedIn’s recruiting and marketing offerings are its main revenue source. LinkedIn Premium subscriptions probably provide nice incremental revenue, but have an inherent — and relatively low — ceiling: most people would only subscribe while they’re job-searching (if at all), then cancel.

So I'm not holding my breath for major changes.

In the meantime, I continue succumbing to LinkedIn Premium's main value proposition: FOMO. But at least I have a strategy now, and try to limit my jobscrolling to keyword job searches for hybrid/on-site roles (i.e. a Pool-Narrowing Factor):

I did just that for the first month of my job search, while in a state of frantic anxiety and LinkedIn-induced FOMO-distraction.

After a personal reset, I came up with a more purposeful approach. Here’s my current concept model for assessing the odds of getting a meaningful first look.

Three caveats:

1. An underlying assumption is you’re actually a good potential match for the job criteria, and not fooling yourself.

2. Another assumption: You're outside the rarified tier that can get a look anywhere (and probably don't even need to apply in the first place).

3. I'm writing from a tech-world perspective; for other industries, YMMV.

Have a Referral-Willing Connection ➞ Odds: Plausible

For a typical job, having a referral-willing connection at the company is the best chance at a meaningful application review.

This could be someone you know directly (a 1st-degree connection, in LinkedIn parlance) who's willing to refer you (formally or informally); or someone who knows someone you know directly (2nd-degree connection), and both of them are willing to refer you.

This is probably an evergreen truth, but particularly the case when job-seekers vastly exceed available positions and/or recruiter bandwidth (i.e., the current tech-world sitch).

The biggest caveat here is companies/organizations where referrals aren't really a thing for normal applicants.

Have a Relevant Pool-Narrowing Factor ➞ Odds: Mid to Long Shot

The next best chance is jobs where you match a dimension that significantly narrows the candidate pool. Pool-Narrowing Factors include:

• Company geography: In-office or hybrid roles will have narrower pools; it also may be easier to find connections at local companies.

• Company size: Small companies tend to be more human, so may be more likely to actually read applications. (On the other hand, they may have ultra-low bandwidth.)

• Specific skillset / experience: It’s a long shot to find roles where you have a meaningfully niche skill or experience, but doing so may increase your chances to mid shot.

All Other Jobs and Scenarios ➞ Odds: Lotto Ticket

In every other scenario, I assume the default odds are infinitesimal.

The wackiest part? <Record scratch> My most-human interlocutor was an AI app.

The experience was unexpectedly profound and clarifying. It illuminated why job interviews can be surprisingly impersonal, and how AI might help bring some humanity back — or at least dispense with the charade.

Practicing job interviews with an AI 

This all happened while practicing for job interviews using Nadia, an AI coaching app from Valence, an employee development platform.

I used the Nadia webapp, and entered this prompt to start the session:

“Hi Nadia, can you help me role-play a job interview? I’d like you to ask me the most common behavioral interview questions. Just ask them one at a time. Give me a chance to respond. Give me feedback on the answer before you ask the next question. Thanks.”

The app asked questions like:

“Describe a situation where you had to provide feedback to someone who wasn’t meeting expectations. How did you approach it? Take a moment to consider your answer, and whenever you’re ready, go ahead and share your experience with me.”

I answered by speaking into my headset microphone. (There’s also a text entry option.) The app recorded my voice, quickly transcribed my answer, then gave feedback. For what it’s worth, while I had previously written and practiced STAR/PAR-style answers, I decided to answer more off-the-cuff rather than trying to recite my pat STAR answers.

To help illustrate why the experience was so profound, I’ll first share a few examples of my answers and the app’s responses. My answers are largely copy-pasted from the app transcriptions; I’ve added some retroactive punctuation, paragraph breaks, and the correct spelling of Viget 🤪.

Example 1

The app asked: “Describe a time when you had to deal with a high-pressure situation. How did you handle it?”

My answer

“So recently, I worked with a cybersecurity startup — a funded startup. I worked directly with the five-person leadership team, most closely with the CEO and technical co-founder. It was a team of two: me and a designer. I was leading.

And the pressure was, we didn’t totally understand the ask; they couldn’t articulate it too well. Neither of us knew anything about cybersecurity. This was my first exposure to it. Some of the team members, some of the leadership team was skeptical we could ramp up on cybersecurity. They were skeptical that we could work with them as technical co-founders.

And so I handled it, first, knowing that I have a depth of experience that I can lean on to be confident that I’ll get through it. So it didn’t feel high-pressure in the moment, though it honestly was.

And so I just followed my process. I did really deep, thorough discovery; talked to every one of their team members one-on-one; did workshops together to understand the business, their problems, ramp up on the domain; did a ton of independent domain research, competitive analysis. And through that, leaned on my skillset and superpower of sense-making to not only kind of say back to them what I heard, but really was able to articulate and synthesize things that they couldn’t — that they still said in fuzzy ways or thought of in fuzzy ways.

So lots of aspects of the industry turned out to be really fuzzy. I was able to give a clearer concept model. I was able to clarify what their real goals are, which was to really reach a new customer segment. And I was able to more clearly articulate the problems they were hoping to solve and give them much more valuable and unique solutions than the kind of surface-level ones that they had initially thought of.

So as a result, we earned their trust. They implemented one of the first of those feature suggestions recently and used it as part of a major partnership announcement with Google.”

The app’s feedback

“That’s an incredibly detailed and robust approach to handling a high-pressure situation, especially in an unfamiliar domain. Your methodical and comprehensive strategy—deep discovery, individual conversations, workshops, and domain research—demonstrates your ability to manage complex challenges effectively. Employing your sensemaking skills to clarify and synthesize information not only helped you understand the client's needs but also contributed to building trust and delivering value.

Your experience underscores the importance of process, research, and communication skills in navigating high-pressure environments. Moreover, the outcome with the leadership team and the implementation of your solutions, leading up to a significant partnership announcement, is a testament to your successful handling of the situation.

Keep this story in your interview toolkit; it’s a solid example of managing pressure and producing results.”

Example 2

The app asked: “Can you describe a situation where you had to make a decision without all the information you needed? Take a moment to consider your answer, and whenever you're ready, go ahead and share your experience with me.”

My answer

“This is actually from a personal project.

So I am a sing-along impresario — I host sing-alongs — and I’ve created my own songbook. And I received feedback earlier this year from three different women who attended that the song keys were not good for them — that many of the songs, it was either too low or too high. Now, I don’t know what the best key is to change them into, but my audience is about 80% women, and so I wanted to just make a better experience for them.

And so what I decided to do is take a number of songs and just put them back into their original key. And I asked the people who gave feedback if they could listen to a few of the original songs, sing along with it, and tell me if the original key worked for them. And they gave me some feedback on that, and it seemed like a good enough option.

And so I went ahead and updated almost all of the songs that had female singers, even though I’m sure the new keys aren’t perfect either. But I just wanted to make the experience broadly better for them, even though I didn’t have all the information.”

The app’s feedback

“What a unique and interesting scenario! Your example perfectly illustrates your willingness to take action to improve your service, even when you don’t have all the information. It’s commendable that you listened to your audience’s feedback and took proactive steps to make the experience more enjoyable for them. By reverting the songs to their original keys and seeking further input, you showed that you value user experience and are adaptable.

One suggestion for improvement could be to discuss how you might measure the success of these changes. For example, you could collect more feedback after the changes or observe participation levels to ensure the modifications had the desired effect.”

Example 3

The app asked: “Can you tell me about a time when you took the initiative to solve a problem before it became a major issue? Reflect on your past experiences, and when you’re ready, I'm here to listen to your story.”

My answer

“Sure. So it was introducing, creating a QA process and structure at Viget for the first time.

So QA was one of the first product management things that I picked up. I was very adept at it and it just made sense to me. It took me years for other things, but I did that from the start. And when I started working at Viget, it was interesting to me that I kept hearing this chatter from devs, front-end devs, “We really need to hire some QA folks and a dedicated role.” And that seemed weird to me because I did QA on every one of my projects. I did it very well. So I wanted to dig into that.

So proactively, I basically did a QA initiative and I met with some of those developers, front end devs, who I had heard that feedback from, met with some of the project and product managers. And my assessment was, no, we didn’t need a dedicated role. The problem was sometimes QA isn’t happening or when it is happening, it’s not being done well. And the reason for that is lack of systems, structure, process, and training. And just people didn’t know about QA or know how to do it that well.

And so from there, I created our first QA process. So I wrote a doc that explained how we think about QA; what are the different buckets of QA — functional, visual, accessibility; which of those we do by default on projects, some performance testing we wouldn’t necessarily do; laid out those expectations. We used that doc internally and we would give it to clients on subsequent projects, so they understood what we would be doing QA-wise. I also created QA spreadsheet templates for a number of different scenarios: browser testing, complex multi-user testing, things like that. And then I did training for the PMs to help ramp them up on that.

And that was about 10 years ago. And that QA process is still the foundation of the current one today.”

The app's feedback

“Taking the initiative to assess and address the QA process at Viget showcases your foresight and problem-solving skills. You didn’t just take the conversation about needing a dedicated QA role at face value; you investigated the issue, gathered insights, and determined that the root cause was a lack of process, structure, and training. Your solution was comprehensive, involving the creation of documentation, templates, and training that not only met the immediate needs but also stood the test of time.

This example speaks volumes about your ability to identify gaps, take ownership, and implement solutions that have lasting impact. Your story also emphasizes the importance of understanding the nuances of your team’s challenges and being proactive in creating efficiencies.

Your approach to solving this problem before it escalated further is exemplary, and it’s the kind of initiative that employers look for in a leader. Be sure to highlight this proactive nature and your capacity to build sustainable systems when discussing your experiences in interviews.”

Why an AI job interview (practice) was so different

I wasn't shouting into the void

One of the most demoralizing aspects of a job search is the constant feeling of shouting into the void. You form a parasocial fantasy relationship with each job listing, imagining that since you applied — and maybe even got a tangential introduction — there’s obviously a real connection this time. Then applications go unanswered. You get ghosted after a final-round interview. Post-interview rejections convey little decision information. You realize the grinding reality of your situation.

Even worse, interviews with actual people often feel like shouting into the void.

Interviewers can’t or don’t engage with your answers (for legitimate reasons). You answer a question; at most they give a short, non-committal acknowledgement that you did, indeed, answer the question; then they ask the next question. You might as well be talking to a — well, I’d say “machine” if that weren’t the whole point of this article. 

As soon as the AI app gave its first feedback, I almost cried. This was the first time during skill or behavioral interviews that I felt like the other person party actually engaged with my answers — that I got any feedback or engagement at all.

My experience felt valued — and validated

As the app gave feedback, I realized this was only the second time in my job search that my experience seemed truly valued and validated.¹

There’s a weird dynamic in many job interviews where the employer is ostensibly already interested and impressed, but they rarely convey that out loud.

Instead, I found myself having to make the case that they should be interested in me, seemingly from scratch. I would think, “Wait, didn’t I get an interview because you’re already interested and impressed? If you’re not, why didn’t you just reject me? Why aren't we digging into the parts of my resume or cover letter that resonated the most — have you even read my resume?” It’s somewhere between gaslighting and playing hard-to-get. (Of course sometimes it’s because … they aren’t interested or impressed 🙁.)

In contrast, the AI coach was actually impressed with my experience!

Consider these feedback excerpts:

That’s an incredibly detailed and robust approach to handling a high-pressure situation, especially in an unfamiliar domain.

This example speaks volumes about your ability to identify gaps, take ownership, and implement solutions that have lasting impact. ... Your approach to solving this problem before it escalated further is exemplary, and it’s the kind of initiative that employers look for in a leader.

It was like: Ohmygodthankyou, finally!!! I generally don’t have imposter syndrome and am quite confident in my past work. (Don’t worry — I have plenty of other self-criticism to deal with.) But this AI feedback made me realize how much I had started doubting myself in the job search. It was so reassuring and needed.

My interview answers felt validated

It wasn’t just my experience that felt validated — I was also getting validation about my interview answers for the first time.

Consider these feedback excerpts:

Keep this story in your interview toolkit; it’s a solid example of managing pressure and producing results.

What a unique and interesting scenario!

The AI coach was impressed with my answers! Not the experiences and skills they reflect, but their effectiveness as job interview stories.

I almost cried a second time when the app called my sing-along answer “unique and interesting.” I’d had that story in my back pocket, but had been nervous about whether to use a non-work — and non-digital-product — example. Now I feel confident to use it in a real interview.

Again, contrast that with the typically stone-faced human interview responses to your skill and behavioral stories. You have no idea if your answer resonated, was what they were looking for, was too vague, too specific, too long, etc.

People are terrible at synthesizing interview answers

The last big difference between an AI job interview and a human interviewer: People are comparatively a joke at synthesizing show-don’t-tell interview answers.

Consider these feedback excerpts from the AI app:

Your example perfectly illustrates your willingness to take action to improve your service, even when you don't have all the information. ... By reverting the songs to their original keys and seeking further input, you showed that you value user experience and are adaptable.

This example speaks volumes about your ability to identify gaps, take ownership, and implement solutions that have lasting impact. Your story also emphasizes the importance of understanding the nuances of your team's challenges and being proactive in creating efficiencies.

There is no way a human interviewer can synthesize a candidate’s answers so thoroughly and insightfully, in written feedback let alone in-real time. Doubly so if a regular employee is conducting the interview, rather than People Team members or other ostensible job interview pros. (But even the latter are likely miles away from the AI app’s synthesis abilities.)

If you’ve ever conducted a job interview, look at the above AI synthesis and honestly evaluate your synthesis in comparison. I have, and my honest answer is: ell oh ell.

AI and the future of story-based job interviews

Based on this experience, if I have to do story-based interviews (i.e., answer skill and behavioral questions), I would 100 percent prefer to do it with an AI rather than a person. (I think people should absolutely still facilitate other kinds of interviews like portfolio shares and exercises / simulations. And people, not AI, should make the decisions throughout.)

Some hypotheses for why an AI interviewer could be better for story-based interviews:

Less demoralizing

An AI interviewer could engage, even provide feedback, without giving inappropriate false hope. (Assuming the app sets appropriate expectations for the candidate.)

Compounded across a candidate’s interviews, this would be a powerful antidote to the demoralizing, shouting-into-the-void aspects of a job search. Save demoralizing for the actual rejection: there’s no need to make the rest of the process feel like an impersonal, ongoing rejection-in-the-making.

Reduce pressure

Story-based job interviews can be uniquely stressful for candidates, in ways that I believe are not an effective proxy for simulating on-the-job pressure.

For one thing, in a job interview you simply aren’t fluent with the material the way you would be when giving a work presentation. After all, nobody perpetually maintains and practices a body of STAR-formatted interview answers about their work experience.

For another, job interviews often carry layers of fraught personal dynamics that aren’t present in job situations. The stakes are existential. You — not your work — are being judged. Your deepest self-criticism and regret are always just below the surface, as you’re forced to reflect on your past and constantly talk about yourself.

I suspect an AI interviewer would mitigate some of those fraught dynamics, even though the existential stakes are ultimately the same. 

10x better synthesis

The feedback excerpts above speak for themselves. If story-based interviews are supposed to assess how well a candidate’s stories demonstrate skills and behaviors, an AI app can articulate that assessment so much better and faster than a person. This would both improve outcomes — employers have much better information to base decisions on — and save employers a ton of time.

Bias mitigation

AI interviews could enable blind auditions at scale. And the 10x Better Synthesis factor would likely identify promising experience that human interviewers might overlook.

Wrapping Up

All of this assumes story-based job interviews are actually useful, predictive, necessary, etc. in the first place. Hot take: I don’t believe they are — but that’s a topic for another post!

In the meantime, if you’re applying to jobs I highly recommend using Nadia or another AI app to practice your job interviews.

Choose a memorable name and URL. Prep that first post. And — natch — use ChatGPT to create a logo!

I'm struggling to pick from so many good options...

Clearity: An instant classic

Here's the prompt I started with:

Make me a logo for my new blog and newsletter, Korr Concepts. I write about fuzzy and funky concept models, and how I make sense of them and turn them into clarity.

Voila:

I have a soft spot for this one. It's peak AIrt uncanny valley: the airbrushed textures; the vaguely disconcerting nonsense-blob elements; the misspellings. "Clearity" makes me laugh every time!

If I were planning to write primarily about AI, I'd probably choose this logo.

Rube Goldbergs: Not bad!

Next I asked:

Can you try a version that's a little more rube goldbergish?

The first try:

Pretty good! I especially like the lollipop whirls and the blue pig nose. But that typo reads as a flaw — not funny enough to use.

I asked for a revision:

Nailed it! Only problem is it may imply a physical-product focus. So I asked:

Could you try a version that incorporates thinking motifs or elements?

Meh — worst of both worlds.

Style Explorations

Next I shifted gears to play with different art styles.

Art Deco

These are both great (Sglpessieng!), but not quite the vibe I'm going for.

Film Noir Poster

Pretty weak: I appreciate the individual noir(ish) motifs, but this is not movie-poster style. (Also what is the man next to the cat holding?!?)

Rock Concert Posters

Next I asked for the style of 1960s rock concert posters:

Hmmm. Not only is this no Milton Glaser — at this point a DALL-E Default house style is emerging: again the airbrush and nonsense blobs; the circles and swirls; the particular color palette and saturation. "Exlainzy and fuzzy finky" can't salvage this one.

DALL-E Default also foiled a stab at 1980s-rock-concert-poster style:

A second attempt is still far from rock concert posters, but at least captures an '80s aesthetic (and apparently the '80s vogue for pom-pommed hand grip strengtheners?):

Polish Cyrk Posters

My wife is fascinated by Polish cyrk (circus) posters (we have two), so I also asked for that style.

Huh. That would have made more sense from the 1960s-concert-poster prompt.

As an AIrt connoisseur, I appreciate this one for the incongruous mix of DALL-E Default style and undercurrent of unease. But not what I'm going for. (Plus it has nothing to do with Polish cyrk posters.)

Polish Resistance Posters

Initially I couldn't remember the term "cyrk," so I asked for the style of "1960s Polish resistance posters" instead.

Wow — that's pretty stunning. Here the nonsense words actually feel like part of the poster: more menacingly runic than unintentionally funny. But overall, it's too authoritarian and alarming.

The (tentative) winner

Now we're talking! The giant eye is right on the edge of authoritarian, but the screenprint effects and a few curves and circles give it enough warmth and humanity. This is also by far the least AIrt-y and ironic of the bunch.

I picked this one for now, but I'd love to hear your favorite(s)!