top of page
  • Writer's pictureJakob Nielsen

UX Roundup: Synchronizing Captions | SUS Survey | First Mac GUI | Student/Beginner Portfolios | Solve the Easy Problem | More UX Songs

Summary: Auto-sync makes YouTube captions easier to create | The SUS questionnaire for user satisfaction is too long | The revolutionary nature of the first Mac GUI | Student or UX Newbie portfolios should solve a real business problem | Solve easy local problems before tackling deep systemic issues | UX songs in 3 more genres


UX Roundup for April 15, 2024. (Midjourney)

YouTube Captions Auto-Sync

It’s long been a firm usability guideline to add captions (subtitles) to videos. They are a big help for non-native speakers, users with hearing problems, listeners in a noisy environment, or users who keep their speakers on mute (maybe because they’re at work).

However, it’s a royal pain to caption videos, so it’s usually not done.

One of the oldest usability findings is that the easier something is to do, the more people will do it. This applies to content creators just as much as it does to regular users. Therefore, I have to congratulate YouTube on a nice advance in its UI for adding captions to uploaded videos.

If you happen to have a transcript of the video at hand, you can simply paste it into the UI, and YouTube will automagically split the transcript up into sections and synch them with the audio track, so that the correct subtitle appears when those words are spoken.

Here’s an example of how I added captions to the song about my sixth usability heuristic, Recognition Rather than Recall. In this case, I did have the lyrics written in advance, so they were an easy paste job. Well done, YouTube!


User interface for synchronizing captions with a video on YouTube. If you already have a transcript of the video, you can upload this text, and the system automatically creates the breakpoints and synchronizes each caption with the time stamps when those words are spoken (or sung) in the video. This task used to be terrible, and now it’s easy, thanks to AI!

The SUS Questionnaire for User Satisfaction Is Too Long

Jeff Sauro and Jim Lewis from the “MeasuringU” consultancy are two of the world’s leading experts on UX metrics. They recently wrote an analysis of the System Usability Scale (SUS), which is a 10-question instrument to ask users about their subjective satisfaction with a user interface design.

SUS is now 40 years old, and the authors asked whether that’s simply too old for use in modern UX projects, listing a variety of recent complaints from other authors. Sauro and Lewis conclude that SUS is not outdated. It’s still a good questionnaire to measure perceived usability.

I mostly agree with Sauro and Lewis. (Of course, they’re the experts.) In particular, it’s only good when something is old and well-proven, because only the best survive that long. (E.g., my 10 heuristics are 30 years old this year.)

However, the opportunity cost of asking 10 questions to find out one thing (users’ subjective satisfaction) is too high. We can only extract so much data from users, and I would rather spend most of that extraction budget on learning more about other things. One question about subjective satisfaction is great. 10 are too much.

This is particularly true since subjective satisfaction is one of the less useful things to know about usability. For redesign, it’s much more important to know why users have trouble with the UI than exactly what they rated it. It’s still good to know whether people like your design, but knowing that they gave a score of 7 on a scale of 1-10 doesn’t give you any more useful information to drive a redesign than knowing, for example, that they rated it 8.

“Why Beats What” for knowing how to improve your design. This is one of my favorite UX slogans. (Ideogram)

Get the score, but don’t waste your extraction budget on extraordinarily detailed discussions of subjective satisfaction.

Interestingly, my recommended survey for user satisfaction is the Single Ease Question (SEQ), which also comes from MeasuringU: “How easy or difficult was it to complete this task?” on a 1-7 scale. Plain and simple and not wasting users’ time. Thus, you can gain more info from your users about more important things where nuanced insights matter more.

Completing long questionnaires drains users. In a usability test session, we only have so many minutes, and we should spend most of them observing real behavior as people attempt tasks with our design. For online questionnaires, excess questions kill the response rate and invalidate the results. (Midjourney)

Lest We Forget: The Revolutionary Nature of the First Mac GUI

Good video on YouTube of Susan Kare demonstrating the Macintosh user interface in 1984. (7 min.) Kare was the designer of the first Mac icons. She’s the icon of icon design! What’s exciting about the video is to reexperience the GUI through fresh eyes, when it was new. Statements like “this is what we call a window,” and “you can write notes on the NotePad even while you are using another application” illustrate the difference between the GUI and the computers people were used to at the time. (Hat tip to Paul Graham for alerting me to this video.)

Of course, we all know that the GUI was not invented by the Macintosh team, but originated in many other places, most prominently at Xerox PARC, and even at the Lisa team within Apple itself. The Mac’s main contribution was that it was the world’s first “cheap” GUI. Price is one of the main determinants for design success.

I put “cheap” in quotes because I remember agonizing for months about whether I could justify buying a Macintosh on an Assistant Professor’s salary, even when augmented by quite fat consulting fees on the side and even though I was eligible for half price through a faculty discount.

However, the cost of a Mac should be compared to other GUIs at the time: my first GUI research project used a PERQ-2 computer that cost more than $10K, so in comparison, the Mac was indeed cheap at $2.5K, and it deserves its credit for making GUIs ubiquitous. And GUI was one of the biggest UI advances the world has seen.

That UI that Susan Kore demoed in 1984 dominated the world of computing for 40 years with really only minor updates, such as color. That shows how groundbreaking it was. Only now is AI giving us a new paradigm for using computers. But as I have said many times, the best usability will really come from a hybrid UI, keeping the best of GUI while adding the AI paradigm. (Adjusting for inflation, these prices correspond to about $30K and $7.5K, respectively, in today’s money for the PERQ-2 and the Mac.)

Moving from endlessly scrolling text to a GUI with windows and things you can click was a major leap forward in computer usability. Unfortunately, we (temporarily) lost much of this with the current generation of chat-based AI. (Midjourney)

Student/Beginner Portfolios: Solve a Real Business Problem

Colton Schweitzer wrote a screed about the many useless design projects featured in most UX portfolios by students or beginners:

  • Solve an irrelevant problem with no business value

  • “Solve” a huge problem with a redesign of a major website or app that has no realistic hope of ever being implemented

Instead, Schweitzer recommends picking a company that already exists and finding problems that they are experiencing. Solve a narrow and specific UX problem that’s costing a company money, rather than showcasing something that will never be implemented in the real world. That way you show hiring managers that you can fit into their existing team and produce work that’s applicable to them.

The job of resumes and portfolios is to survive the first ruthless screening by an overworked and hard-nosed UX manager who’s looking for people who slot directly into his or her existing team and start making money for the company. Therefore, you should showcase realistic business designs in your portfolio. (Ideogram)

Don’t Boil the Ocean: Solve the Easy Problem First

David Hamill recently noted that it’s usually not feasible to propose wholescale technical overhauls in order to resolve an interaction issue. Instead, simple and local fixes are much more likely to be implemented, even if they don’t fully resolve an underlying systemic issue.

Whenever a pragmatist suggests a simple solution, you will often find theorists getting on the high horse and proclaiming that the fix only soothes the symptom but doesn’t go to the core of the problem. This may be true, and there’s some value in conducting a root cause analysis in case one discovers a deeper problem that’s also not too difficult to solve.

But in the real world of actual UX design (as opposed to academic discussions of the full ecosystem), helping users (and the business) here and now usually comes first. The foundation of my discount usability ideology has been the same for 35 years: the perfect is the enemy of the good. This is a quote from Voltaire in 1770, so it’s rather a long-lasting insight.

Hamill uses the beloved Silicon Valley simile of “boiling the ocean” to characterize the high-horse desire for big-picture solutions. I had a hard time getting Midjourney to depict the boiling of the ocean, probably because it’s impossible to do.

Being of Viking heritage, I thought of a more visually appealing metaphor for attempting the impossible: once Thor, the God of Thunder, went fishing and got Jörmungandr on the hook. He tried to reel in his catch but failed. Later in the myth, Thor realizes that Jörmungandr is the serpent that encircles the entire globe, meaning that even a god can’t pull him out of the water. (Also Midjourney.)

If Thor can’t do something, you probably can’t either. Focus on the achievable, even if elitist theorists belittle you.

UX Songs in 3 More Genres

I’m happy to report that the new generative-music service Udio has improved its response times dramatically since I wrote about it last Friday. Getting a song is at least 10 times faster now.

Because the tool is faster, users will attempt more variations of their songs. I made 3 new songs about my usability heuristic number 6, Recognition Rather than Recall. I used the same lyrics for all the new genres, to make it easier to compare the composition and performances.

Which version do you prefer? Let me know in the comments.

I attempted a 4th genre: classic 1780s opera, but the result was so atrocious that I won’t inflict it upon your ears. AI is clearly not up to composing classical music yet. Not that I expect Mozart from an AI for the next 10 years or so, but classical music is an important enough genre that they need to cover it.

Udio also added a feature to generate cover artwork for the songs, but these images are dramatically worse than what we get from dedicated image tools like Ideogram or Midjourney. Unfortunately, they don’t allow importing images, which seems a major oversight.

Jazz concert with AI songs about Jakob Nielsen’s 10 usability heuristics. (Ideogram)

Additional musical genres in 20 seconds: the 10 Heuristics are also now a fabulous Broadway Musical! (Ideogram)


Top Past Articles
bottom of page