The SaaS counterculture

And the SaaS status symbol.

America’s original maverick.

Because I could not stop for Death –
MKBHD kindly stopped for me –
The Review held but just Ourselves –
And Immortality.

—

Unfortunately for the Humane Ai Pin and the rabbit r1, it wasn’t the good sort of immortality. Three weeks ago, Marques Brownlee, a wildly popular YouTuber who goes by MKBHD, reviewed the Humane Pin, a portable AI device that’s supposed to help you with tasks throughout your day. He did not like it. “This thing is bad at almost everything it does, basically all of the time,” he said. “Nobody should buy this device right now.” It’s “the worst product I’ve ever reviewed.”

The video went viral, until it was outdone by this week’s review of the rabbit r1. According to MKBHD, the r1, a portable AI device that’s supposed to help you with tasks throughout your day, “has a lot in common with the Humane Ai Pin”—mainly, that it “is also bad at a lot of stuff.” Like the Humane Pin, it has limited capabilities, frustrating interfaces, and its core feature—answering questions about what’s around you—is slow and unreliable. It’s an alpha product, he said, sold at general availability prices. 

Towards the end of the review, MKBHD asked a bigger question about these AI products: “What are we doing here?”¹

A lot of these tech companies are developing tech kinda backwards. They're delivering such unfinished products, that it actually makes them nearly impossible to review.

Like, it feels like it used to just be, make the thing, and then put it on sale. Now, it's like, put it on sale, and then deliver the half-baked thing, and then iterate and make it better, and hopefully with enough updates, then it's ready, and it's what we promised way back when we first started selling it. And then this whole period in the middle is a mess.

He goes on to say it’s happening with other, more mature hardware products too. There’s “one major feature of every smartphone launch that gets announced, but that's not coming until later in the year.” And cars, which now run tons of software as well, are getting delivered in “a half-finished state, where you just don't get a lot of the features that you paid for.”

A bit later, he speculates on what’s causing this trend: AI products need data to train their models.² Though these devices are hardware products, they’re increasingly differentiated by the software that powers them—i.e, the capabilities of their AI agents—and not the technology in the physical machine that you buy. Most of this software runs entirely in the cloud.³ Moreover, whatever software that runs on the device itself, like its operating system or any local AI models, can be regularly refreshed via downloadable updates. Nikon couldn’t ship a busted D50 digital SLR camera in 2005, because the consumer had to use what was in the box, and only what was in the box. But today, Humane and rabbit can ship a busted camera, and fix it in post. 

Which is, of course, exactly how SaaS products also work. 

In other words, the new delivery model for hardware tech—a relatively simple box, and complex software served remotely—is the same as just about every modern web application. It's just that in the latter case, the box is a Chrome browser rather than a nuclear orange Game Boy. And if that model is encouraging hardware makers to ship underdone products, are the incentives not the same for SaaS vendors? 

Or, more bluntly—while we're on the topic of questioning the entire SaaS business model—are SaaS products worse than the software we bought in a box?

Having worked with a lot of SaaS companies, I can at least say that it's not a totally unreasonable question. Most SaaS vendors are relatively lax about bugs, and use their customers as their primary quality assurance teams.⁴ It’s probably a rational decision, too—why spend weeks testing for every possible bug and rough edge when you can crowdsource that work to your users, and, in theory, quickly fix whatever issues they find? Though you could argue that the customer won't like paying you to be any your unwitting guinea pig—and they won't—they’ll probably tolerate it, especially as this sort of practice becomes the norm. Vendors ship incomplete software, the threshold for what makes a bad release slips a bit further, and the whole thing spirals in a slow devolutionary loop, until people are blaming reviewers for not liking useless garbage, because of some open-ended promise that it might one day not be useless. 

All together, it’s a bad pattern for both customers and companies. Customers get shoddy software. And companies get stuck on an unsustainable financial treadmill, where they have to keep building just to keep up with what they’ve sold. For SaaS businesses, that keeps them from ever operating in “maturity mode”—i.e., the period when they actually make money. 

Which makes me wonder—will the SaaS model be as enduring as we all seem to assume that it will be? The financial math might not add up. The prevalence of the subscription services—from B2B vendors, from content providers, from individuals who want to monetize their personal brands through newsletters, Twitch channels, and fan accounts—is creating a lot of subscription fatigue.⁵ And if SaaS vendors are incentivized to make lower quality, less complete products than software-in-a-box vendors, it’s inevitable that those incentives will eventually trickle into the software that gets shipped to customers. 

None of which is to say SaaS is bad, or doomed, or whatever. It just seems that, at some point, the fad will go through some sort of countercyclical contraction, if for no other reason than companies will try to differentiate themselves by being weird. There are already some cracks. Monthly subscription fees are slowly shifting towards metered, usage-based fees.⁶ 37signals, a company that tries to differentiate itself by being weird, recently introduced an option to buy—not lease, buy—their software with a one-time fee. 

If I were thinking about starting another software company, I’d be tempted to do the same.⁷ Build something that you don’t have to keep building forever; sell the product based on what it is today and not based on what you can promise it’ll be in the future. If MKBHD is right, there’s a lot of appeal in that, and a market for it that’s waiting to be sold to.

Ultimately, that’s the interesting question here: Will someone pull this off? Will someone build a big software company that isn’t a SaaS product?⁸ Will there be another Salesforce that does to the cloud what Salesforce did to self-hosted, perpetual license software? And if there is, how many other new companies will decide that that’s the new thing to copy?

I don’t know. But as a software customer who doesn’t want to buy stuff based on roadmaps and upcoming releases, I hope so. Because hope isn’t optimism. Hope is a thing with features.

---

Status software 

Speaking of things with features, back in 2008, a software developer released an iPhone app called I Am Rich. When you opened the app, it showed you a picture of a glowing red gem. There was a single button below the gem; when you tapped it, the app told you this:

I am rich
I deserv it
I am good,
healthy & successful

It cost $999.99.

In the App Store, I Am Rich described itself as a “work of art with no hidden function at all.” In a review, the New York Times described the app as a “downloadable program that promised to do nothing except signal to the world that its buyer was wealthy enough to have spent $1,000 to download an image of a multifaceted ruby.” A day after the app was released—and after eight people bought it—Apple removed the app from the App Store, and said nothing at all.⁹ 

Ok, I have some questions:

• If the red gem in the app was an NFT that got minted the moment you purchased the app, would people buy it? Would Apple ban it?

• If the app had marginally more functionality—you could customize your gem; the app included a basic calculator; it gave you a personality test and wrote you an inspirational limerick; whatever—would people buy it? Would Apple ban it? 

• If the app had marginally more functionality and felt meticulously crafted—the app was a clock that plays The Godfather theme every hour; it was named after a famous actress; the app was made in 2001 and has been aging ever since—would people buy it? Would Apple ban it? 

I have no idea what the answer is to any of these questions. It seems like Apple wouldn’t ban the NFT factory,¹⁰ and people might actually buy it?¹¹ But it seems like Apple probably would ban both the overpriced calculator and the cask-aged clock, for the same undisclosed reason it banned I Am Rich? And, even if Apple approved those two apps, I doubt people would buy them?

Which is all at least a little weird! The first app does nothing; it’s I Am Rich, built for the laser-eyed crypto goons. The second app is just the first app, but it does something—and somehow, that functionality seems like it would make it less appealing and more of a gimmick. And the third app is basically an exclusive luxury good, which is an exceedingly normal thing for people to want. If Rolex can produce a million watches a year and sell them at a huge premium as a status symbol, why can’t some software maker do the same?¹²

Put differently, on this last point, why isn’t there status software?

It could be that software doesn’t cost anything to manufacture on the margin, and people won’t pay luxury prices for something that’s easy to make. But perfume is cheap to produce. Plus, developing software is very expensive, even, in some ways, on the margin. 

It could be that software isn’t scarce—if everyone can own a thing, nobody will feel special for owning it. But that doesn’t really work either. Software could be scarce, either by choice or through some blockchain hocus pocus. Software isn’t scarce because its makers don’t want it to be, which begs the question as to why that is.

It could be that people only want to pay luxury prices for craftsmanship, like a wearable three-burner gas grill.¹³ Loro Piana enthusiasts want to dress like Kendall Roy and to ensconce themselves in an “extraordinary blend of two cashmere fabrics fused together…that is not only luxurious, but exceptionally rich and enveloping.” But that makes the lack of status software even more mysterious. Software is, quite literally, meticulously engineered. Most applications have far more than 57 complications. And if that’s not enough—or if the software is still “too cheap” to manufacture for it to qualify as a luxury good—just add more.¹⁴ 

It could be that those additional complications aren’t useful; expensive stuff needs to be useful. But I mean, please.

It could be that software isn’t conspicuous enough. Status symbols only work if they’re visible; how do you show off your Balenciaga web browser, or your Tiffany to-do list? First, there are lots of ways to tell other people what software you’re using. Second, most luxury goods exist solely to be status symbols. Necklaces weren’t practical things that eventually became status symbols; people created necklaces to be conspicuous. If we want to show off the software we bought, we don’t have to make our note-taking app public; we could just create some other “category” of software that was designed for other people to see. 

It could be that software just isn’t cool. We don’t brag about orthodontia brands, because orthodontia is embarrassing. Perhaps software is the same. This seems closer, though it still seems incomplete—why isn’t software cool? There’s nothing awkward about admitting you use a computer. If having the right water bottle can make you cool, why can’t some app on the phone that we stare at for seven hours a day be cool too?

It could be because software makers aren’t cool.¹⁵ Computers used to be nerd stuff. Though people have favorable opinions about today’s tech companies, they haven’t fully shed that legacy. This seems like the most plausible explanation. Because of software’s origins, software as a status symbol had a cold start problem: It wasn’t originally cool, so we never made it scarce, so it never became cool. The nerdy brand evolved into cheery egalitarian one instead of edgy detached one, so efforts to make it cool felt forced. And so, even though there’s nothing inherently weird about a celebrity endorsing IBM Watson, it feels weird when Serena Williams endorses IBM Watson.

But could that change? If someone built an expensive photo-organizing app, paid Billie Eilish to use it but not shill for it, and arranged some paparazzi photos of her scrolling through it, could you astroturf a luxury software market?¹⁶ It works for shorts.¹⁷ Or, what if instead of a bunch of software engineers trying to make a cool app, a “cool” company tried to make a cool app? LVMH makes thousands of products; why not make a calendar app too? 

Maybe that’s where the real answer to all of these questions are: Inside LVMH’s boardroom. They have staggering amounts of money. If they wanted to make software, they could. If they wanted to hire some designers and engineers, they have tons of perks that they could offer that Google and Facebook couldn’t. So why haven’t they? They know status symbols better than anyone else in the world. Why can’t software be one?

1. Shoutout to Derek Steer for sending this to me. 

2. From the video:

Look back at the beginning of when Tesla started shipping cars with all the autopilot stuff. They were selling cars, and they got popular for other reasons. It was a good electric car with good range, and a good charging network.

And then people started sort of beta testing this autopilot stuff, and Tesla starts collecting millions of miles and hours of data, from how it works in the real world. And so all of that data helped their systems learn, and get the head start that they need, on trying to make the best autopilot software.

So the Rabbit, I'm sure, would like to use their price advantage [it costs 200 dollars, compared to the 700 dollar Humane Pin], and get as many of them out into the real world as possible, into people's hands, so they can start beta testing, and having it trained and getting good at a variety of tasks.

3. This was one of MKBHD’s big complaints about these devices. They’re very slow, because when you ask it a question or take a picture of something you want it to identify, that data is sent to and interpreted in the cloud, rather than on the device itself. 

4. Not all SaaS vendors! I’m sure someone will email me and say that their company is passionate about quality, and that they are customer-obsessed. Which, sure. But also, it’s undeniable that the SaaS model at least lowers the incentive for software vendors to worry about quality. And as Charlie Munger said, show me the incentive and I'll show you the outcome.

5. Counterpoint: The youths don’t mind. 

6. For example, most early SaaS products charged a per month fee for every user who had an account on their service. Then, companies (popularized by Slack, I think) started charging only for users who used the product. And now, SaaS companies are moving towards volume-based pricing, where you pay for every row of data you sync, table you monitor, or second of compute you use.

7. I mean, no, if I were thinking about starting another software company, I would get myself lobotomized.

8. To be clear, my hypothetical company won’t be this company, because nobody, me most of all, wants there to be such a thing.

9. If you want to participate in a performance art project, there is no app for that. 

10. If nothing else, I doubt Apple would ban an app that was free to download and did nothing but offer you a bespoke NFT to buy for $999.99. Which is functionally the same thing as a $999.99 app that generates an NFT, which is functionally the same thing as I Am Rich.

11. Or at least, people living in 2021 might’ve bought it.

12. Maybe some do? Some invite-only products like Superhuman, Clubhouse, Ello, and Airchat were exclusive when they first came out, though they weren’t lasting status symbols. Perhaps, however, that’s how software status works. It’s more like the food scene than the fashion scene. You don’t show off by consistently using some enduring high-status brand; you show off by always having access to the hot new thing before anyone else does.

13. Or is it a wearable Hot Wheels car? I don’t know. 

14. There is, of course, expensive software. Usually, however, expensive software is expensive because it does more and works better than cheap software. I’m not aware of any software that’s expensive because people want to show off its logo in their browser tabs when they share their screens on calls.

15. Like, I make software.

16. “My baby want a Box subscription.”

17. Did it happen for Merriam-Webster??? Does Kendrick Lamar actually use Merriam-Webster over Dictonary.com?? Did Merriam-Webster pay for this?? Did it work?? I have so many questions.