TWA Hotel at JFK

Last Saturday, during Labor Day weekend, Lisa and I booked a couple hours at the pool at the TWA Hotel. We took the A to the AirTran, which was fine, if not a bit laborious. Getting to JFK is always 50% more hassle than you’d expect. The travel to and from would have been much better with a car, but we don’t have a car and taxis are ~$100 each way. So, A train. It’s fine!

The Saarinen-designed space is incredible, an airline terminal that was in use for around 50 years. I felt so comfortable there, and in literal awe. It’s truly beautiful. Architected spaces could and should feel this way everywhere, but we’ve subordinated the beauty of public places to the profit motive. It’s a shame. Every wave of enjoyment I felt from the Saarinen terminal came with a secondary wave of sorrow. I wish we still prioritized the attractiveness of the places we share as communities. America would be better for it.

Another shame is why this beautiful terminal was abandoned: it was never designed for our post 9/11 TSA security needs. The building was built to accommodate flyers checking in and getting to their planes. There is no room for long, snaking lines and X-ray machines. So it goes.

I took a bunch of photos, but by no means captured it all. Photography isn’t the medium for appreciating the building, anyway. You have to visit to get it. Highly recommended! Plus, the pool is nice.

The National at MSG

The National last night at MSG. Always a great show. The new songs were actually the highlights for me, surprisingly. The guitars sounded so great in them.

Can't believe it's been 10 years since Lisa and I saw them last at Barclays in Brooklyn, June 2013.

AI, Art, and Actuality

Derek Thompson responds to the momentary popularity of an AI generated song featuring "Drake" and "The Weeknd":

Some observers look things in a dystopian direction. It didn't take much to imagine a near future where fake songs and real songs intermingled, where, for every authentic Taylor Swift track, the internet was replete with hundreds, thousands, even millions of plausible Taylor Swift knockoffs. Inundated by AI, pop culture would descend into a disinformation hellscape.

And comes to the conclusion I did with one of my recent posts. He continues:

[L]ately I've become a little bored by the utopia-dystopia dichotomy of the AI debate. What if writing a song and dubbing in celebrity voices doesn't clearly point us toward a disinformation hellscape or a heaven of music-writing creativity? What if the ability to send media that make you sound like a celebrity to your friends is, fundamentally, just kind of neat?

AI-generated art is going to distract people for awhile. The novelty of it will leave less time for audiences to appreciate living artist's work. More great art allowed to fall out of sight. We'll encounter dozens of things like AI versions of Beach Boys songs sung by Paul McCartney, for example. Neat. But forgettable. These things will eventually descend to their natural place at the bottom of cultural importance. Never entirely gone, but also never more than brief curiosities.

This isn't an argument against AI's effect on other forms of labor. In those fields, there will be massive waves of new efficiencies. Art, however, is human first. Any technology, new or old, is only a tool to connect human minds to other human minds. AI, as it relates to art, is only another one of those tools.

And you can't go to an AI-generated concert performance. When Albarn first launched Gorillaz, he tried to perform while hiding behind a screen on which the cartoon band was projected. Fun, for a moment. He quickly dropped the whole cute conceit and played on stage like any other group. In the end, being human is what counts.

Tidbyt

A month ago, I bought a Tidbyt. It's a fun, beautiful little lo-fi display screen for displaying micro content. The thing is really well designed and looks great on the shelf. I have six apps in rotation, so there's also a nice variety in what's displayed. A 3-day weather forecast. A photo of Lisa and I. A pretty Day/Night globe view. A fuzzy clock ("Twenty-Five Till One"), a day/month/year progress thing, and an MTA train tracker for the F line. There are a ton more apps, but these are the ones working for me at the moment. Being a relatively new product, I'm sure more great apps will be added as time goes on. I'd love to design and build one myself if I can think of something new to make.

On mornings when I go into the midtown office, I check the F train time to see if there were any delays. Yesterday, there was! Imagine that. Signal malfunctions on the line meant the train ended up stuck on the track for a long while. So I knew to walk to Jay Street to catch the A/C instead. It's a silly little techy box, but I love it. Highly recommend buying one for yourself or someone else.

Anatomy of a Scam Call

I just received a "scam likely" call from the Netherlands. First time I remember getting spam calls from outside the country, and my parents just so happen to be in Amsterdam today. Provides a super weird clue to how scam callers designate who gets called and from what numbers. I knew spammers have email contact lists from purchased/stolen data lists. I hadn't considered that phone spam calls would have the same kind of origin.

What must be happening here is that one of my parents' numbers are in a spammer's list, and the spammers know where they're currently located. Creepy. Sending me a phone call from the Netherlands today is supposed to catch me worried that something has happened to my parents, that I would instinctively answer out of concern.

Most of my scam calls come from Indiana. Before now, I thought that was because I have an Indiana area code for this long-held phone number. Now I suspect those calls are somehow attached to friend or family who have been caught in a data breach somewhere.

I'm also assuming this was automated, and not a person manually cross checking data tables. Scam artists are clever and always improving their methods. Automation. Location. Batch calling. I knew better than to answer this call, but I can't say the same for any of the older folks in my parent's contact list. One of my senior family friends was recently scammed out of a thousand dollars or so. I wish I could do more to protect the people I know. There's only so much you can do.

The Future of AI and Machine Learning

There is a ton of chatter right now about artificial intelligence and machine learning. The tech behind it is advancing really quickly. People are wondering when, not if, it will replace human time and effort at accomplishing tasks. Particularly in the creative spaces: coding, image creation, technical and creative writing, etc.

I have a nagging sense this is a huge to-do about nothing. That AI & ML is this growing ball of yarn which seems to be encompassing everything, getting all mixed in to our lives, jobs, way of life. But then, if we take a wide step to the side, we'll see it's not as integrated into our lives as we thought. We could snip a single string and the whole thing falls away.

People will use AI to generate art for awhile until the look and style of those images runs out of novelty. Humans crave creative novelty. It's in our DNA to always be searching for the new and different. I foresee a decade or so of people finding ever more interesting ways to use AI imaging until ... we culturally decide, en masse, that it's time for a change.

We do this all time in every form of art. Push the envelope forward into new territory for awhile, then retreat to the familiar. Then push forward again. Then retreat. Like in music where the pendulum is forever swinging back and forth from high production to low, from expensive to democratic, from dance to bedroom, etc. Our exploration with AI is surely going to incorporate the same pattern. We'll have fun with it for a time, but then come to understand exactly where the human touch is missing. At which point, we'll prize human made art again.

In an article called The Age of Average, Alex Murrell writes about this phenomenon where every culture in the world essentially wants the same thing, on average, in their ideal art. "Despite soliciting the opinions of over 11,000 people, from 11 different countries, each of the paintings looked almost exactly the same," Murrell writes. He quotes the artist who did this research: "In nearly every country all people really wanted was a landscape with a few figures around, animals in the foreground, mainly blue."

I predict that we push and play with AI for a number of years — and having a good deal of fun and excitement with it — up until the point where every image it produces feels to us like the one billionth landscape with a few figures around, animals in the foreground, mainly blue. AI won't have made artists redundant. It will remind us why we always needed them to begin with.

In so much of what I read about AI, folks are forgetting that it's only a tool. There's a magic in it, but the magicians understand the tricks. Yes, most of us will be fooled for a long time, only a few people really understanding the technology behind it. Eventually, the magic will look more like science and take on the aura of normalcy. We'll get it. We'll outgrow it.

We're in the early days of experimentation. Tons of cool stuff is going to be made and will blow our minds. Lots of AI stuff will be amusing. Some of it will really help us perform better at our jobs. But, some of it will be poorly thought out, or indistinguishable from a dumb April Fools joke. Like this new Software as a Service called SyntheticUsers which purports to be a user research tool that doesn't have real users. Is it even real? Is it a joke? It doesn't matter which to me, because it's an obvious swing-and-a-miss at solving a real problem *, and so it's a joke either way.

What I'm pushing back on, I see now, is the hyperbole. Artificial intelligence will in fact change our lives. To what degree that's true is probably less than the chattering class is promising. I'd say on the level of the invention of plastics, and not so much the discovery of electricity. It's a new type of thing to keep our leftover pasta in. Not an entirely new way to absorb nutrients from food. I'm excited people are excited! But my expectations are tempered.

* For the non-tech folk reading this, I can explain the stupidity. The point of user research is to discover the unexpected ways real people misunderstand or misuse your products. You need living human beings to test this out. Any machine learning AI would only be reproducing common mistakes SyntheticUsers have fed to the algorithm. It will forever miss the wonderfully clever and stupid ways people behave. What you would learn from SyntheticUsers would not be how real people approach your software, but how AI approaches your software. It's so perfectly pointless, I have to assume it really is a joke.

Planting on the Roof Deck

On Sunday, Lisa and I planted a mix of plants in the boxes left for us on our roof deck. It was the first time for us doing any kind of gardening for ourselves and it went pretty great. I'm really happy with how it came out. The boxes could have used another couple inches worth of soil, which she pushed for but I wasn't sure about. I thought the mass of roots and soil which came with the new plants would make up the difference, but I was wrong. She was right. I'll probably buy some more soil soon to raise everything a bit.

I watched a bunch of YouTube videos trying to learn as much as I could about rooftop gardening, and gardening in general. A couple tips were really helpful. Originally, I had only thought about planting a single type of flower, maybe two. I hadn't considered mixing grasses, herbs, and bushes with flowering plants. That's known as polyculture, which promotes healthier soil. It's also more attractive! And it introduces another level of design into the project. I'm really looking forward to thinking through the arrangement more fully next year.

I also learned that for a windier location like our roof deck — and it does get really windy, sometimes — it's a good idea to plant bushes or taller grasses which shield smaller plants from the gusts. I've done that a little with a boxwood bush in the western-most corner. Not yet sure that's enough. We'll see.

We only shopped at the Lowe's in Gowanus, which didn't have as much variety as we hoped. The didn't have any ornamental grasses, for instance, which I wanted to incorporate. We did come home with a good variety, though. Next year, I'll do more shopping around to get what I want instead of only what's available in one shop. Here's what we planted: yellow daffodil, mediterranean pink heather, english daisy, grace ward lithodora, early bird radiance dianthus, begonia, and green mountain boxwood. Also some mint, rosemary, and basil.

Blogging About Blogging

It's so great to see people returning to blogging after so many years. The mass exodus from Twitter has led, predictably, to folks reimagining how they read and write on the internet. Before Twitter and Facebook honey trapped everyone's attention, blogs had been one of the main places of choice to post. It's possible, though not assured, that they could make a comeback.

My community online is made up mostly of designers, developers, and others working on software products. For reasons that might seem obvious, this group of people tend to be in the first wave of any mass movement happening online. The people who build the web tend to know the web best. Generally, at least. They're all testing out blogging again, one by one by one. My "People" folder in my RSS reader is alive in a way it hasn't been in years. Personal websites are coming back to life.

I'm watching for two outcomes from all of this movement. One: will all of this design and engineering attention returning to blogs, RSS, and other forms of web publishing result in new innovation? Better newsreaders? More creative uses of RSS? Clever applications of ActivityPub? And Two: will the non-tech community follow us into the next phase of web publishing? That will depend on how easy we can make it for people to start their own blogs, or to join federated servers, etc. It's a real challenge, as it depends as much on social dynamics as it does tech innovation.

One piece of the puzzle still missing is a centralized place to find citizens of the internet. And for good reason. Decentralization is so important to the concept of the open web. Despite needing to avoid any single institution controlling that list, it would be hugely beneficial to have one place to go to find where your favorite people are publishing. That's what my product Feeeds was supposed to solve. It was a way for a person to curate all the RSS feeds they produce online into one bundled feed. Perhaps this is a product which could be federated, too? Could this feature be added to Mastodon?

But maybe the best web possible is one that embraces the chaos. It's possible that any centralization of a certain strength tips into producing more negative outcomes than positive. I'm enjoying the conversation that's happening now. The death of Twitter has revitalized interest in all of this. The future of the web seems open in a way it didn't six months ago. It feels like nearly anything can happen now. The exploration will be well documented on the blogs.

GBV, Terminal 5

Guided By Voices sounded pretty great last night. They opened for Dinosaur Jr., who were also good, but I guess were the closer because they’re louder. But got to take my friend Brian to his first GBV show. Doing my damndest to turn him into a super fan like me.

Breaking Figma Updates

I've been getting a lot of Figma library updates in my files which weren't triggered by human action. Neither I nor anyone on my team made updates to the libraries, but yet components show 'Updates' which must be accepted — or ignored forever until the end of time. What's causing this? Ghosts in the machine?

I have a guess. It has to do with failing to perfectly use Figma's every feature exactly as intended. Figma is far more technical a tool than they'd like to admit. If you want to use any of their smarter features, you're stepping outside of Design work and entering Engineering Land. The smallest error means your rocket might explode on its way to space.

Figma has a ton of great auto-layout and library features. They're great and they save time and reduce inconsistencies. The catch is that you have to use those features exactly how Figma wants you to. Figma expects you to have read and memorized every line of their documentation and watched every instructional video before you've added a single object to a file. If you start working without fully understanding what a Variant State or Mode is, or how a Hugged Content group with 8px of horizontal padding will behave when stretched, well, you're likely to have broken comps in the near future.

What breaks the comps? Figma updates.

One of my design system components which keeps getting these required ghost updates is my Tab Group. It's a component with an auto-layout group inside an auto-layout group inside an auto-layout group inside an auto-layout group inside an auto-layout group ... with a few variants. And this is a basic UI module. It's just a row of tabs. But multiple times now, this component has for unknown reasons been triggered to update, and after updating, options for it in my comps turn on and margins are broken, etc.

Here's what happened. Figma released an update which subtly changed the rules for which a 'hugged' group behaves when grouped with sibling auto-layout objects that's also inside an auto-layout group which has a variant. Or whatever. Something to that effect, something extremely specific. And it broke my component. The point is that their very explicit rules (likely not clearly written or documented) are so fine-tuned that there is zero curtesy afforded to the user. A component might accidentally be working how the user intended, yet when Figma updates their rules, the component which had been looking great is suddenly upside-down and backwards.

Engineers are used to this. It's a standard part of writing code. Code merges can sometimes have unintended consequences. This kind of thing now comes for designers. The smarter our design tools get, and the more they allow us to do with clever features, the more they're going to break in frustrating, time-consuming, comp-breaking ways.

Figma is great. Sketch is great. I'm really appreciative of both of these tools and how they help me do my job. My takeaway is that I need to focus on the comps I'm currently working on and care less about keeping old comps at 100% fidelity. Design for the current sprint. Design for the current quarter of backlog work. But once those designs are in production, accept that those old comps will slowly rot. Figma will break them as their app changes. Which is fine! Design work for software and the web is ephemeral. It's made easier with great tools, but those tools have no loyalty to you or to the integrity of your designs. They have an app to maintain, and if their updates break your old designs, that's not their concern.

So go ahead and accept those updates, comp-breaks and all. You're already onto the next thing anyway.