maraoz.comGreat Software
I.
A year ago, I wrote about my nostalgia for the olden days of the internet, and I discussed my understanding of why all apps enshittify and wondered if new ways of building software could solve the problem.
I often complain about how bad most modern software is sharpware: buggy, disrespectful of users’ time, attention-grabbing, and privacy-awful. I’ve long been trying to understand why most software sucks so much.
Let’s now try the inverse thought experiment: what makes great software?
II.
We’re spending more and more time using devices. Worldwide average is 6 hours and 40 minutes per day. I myself spend ~8 hours per day between my laptop and my phone. It might be time to ask more of our software, right? That raises the question: what would really good software look like?
A weirder way to put it: what does $5,000/yr software look like? What would a piece of software have to be like for someone to be willing to pay $5,000 per year to use it?
So, on a long walk, I started day-dreaming what this kind of Platonic software looks like. I came up with a list of requirements, in rough order of importance:
0 - No ads (duh).
1 - No notifications, ever.
2 - Never ever sends me an email, unless I explicitly ask it to (eg: I configure an email alert, or I hit “forget password”). No “thanks for joining XYZ” email, no “5 tips for making the most out of XYZ”, no auto-subscribe to the marketing newsletter.
3 - It never ever tries to sell me anything again. If I’m already paying a yearly subscription fee, I don’t want it to offer me “upgrades” (I’ve paid for so many apps, only to then suffer awkward upselling… Dropbox, I’m looking at you)
4 - My data is never sold. The app makes money from the yearly fee, not from my data. I’m treated like an actual valued customer (care, privacy, deference), not like a “user”.
5 - It allows me to turn off any and all features I don’t like. No new features are forced on me (YouTube shorts, I’m looking at you).
6 - The developers improve it slowly, and focus on polishing existing functionality to an obsessive degree before shipping new stuff.
7 - Premium human-only support, ideally by the founders/developers. No 3-day wait, no soulless form-filling, no copypasta responses without reading, no AI assistants.
8 - It’s fucking FAST. It works amazing when offline. No triple-loader-spinner into splash-screen into lazy loading components into loading all the app from the network… (banking apps, I’m looking at you…)
9 - Created with a quality mindset, not a growth mindset. Planning is made in decades, not quarters. If you charge thousands of $ to your users, you don’t need to grow 25% monthly or something. This destroys the tyranny of the marginal user effect.
Would you pay $5,000/yr for an app that promises this, and actually delivers? I wouldn’t. This all sounds amazing, but $5k is a lot of money.
III.
What about $1,500/year for all that? I would, and I actually do. Kubera is $2,500/yr and I pay it happily. Xapo Bank charges an annual fee of $1,000 and it’s the best personal bank I ever used. I’d also pay $1,500/yr for Mercury, no questions asked.
I wish the world had more of these. It’s hard to quantify how many human hours (and frustration) could be saved each week by well-designed, well-programmed software, but my guess is that it’s quite a lot, and the problem is getting worse.
However, I think we can do even better than the list above. I believe software can be cozy, calm, enduring, uplifting, luxurious, austere, evergreen, handcrafted, wise, awe-full…
After all, we created all that bad software. We can just decide to create something different. Something we dare call great software.
IV.
10 - No bugs, ever. If I hit a bug, they refund me the yearly fee.
11 - A promise (or even better, a legal structure) that ensures the software will continue to exist 50 years from now, and will retain its core values.
12 - Made by the best software artisans. Only God-tier programmers, designers and product managers work on it. No junior devs, no AI slop, only the best of the best artisanal human craftsmanship. Also, made by a tiny team of at most 5 people.
13 - Privacy policy and Terms of Service are readable by a normal human, or at least there’s an honest non-legalese summary.
14 - No telemetry. Stats are for growth-hacking, which this kind of software doesn’t want or need. If users have feedback, they can contact support (which is amazing, see point 7).
15 - No dark patterns.
16 - The design is long-term stable. It is based on timeless principles, not UI fashion. Features are never deprecated. The visual system is locked in five-year cycles. App redesigns are opt-out-able.
17 - Very generous bug bounty and public security incident log.
18 - Minimized external dependencies - sooner or later, they will inevitably leak their wrong values to your app.
19 - Intentionally throttled growth.
V.
Is this all even possible? I think so. Please, build such software: it will be great software. I’ll pay for it, I’ll invest in it. The world will be a less frustrating, prettier, calmer place.
The standard is pretty damn high, but then again we’ve only been running software since 1948… It’s been 77 years of software voraciously devouring the world. Only 77 years…
We might not yet know how to build truly great software.
Final Words
If you liked what you read, or would be interested in using such great software, consider subscribing below.
Further Reading
Acknowledgments
Cover photo by MidJourney.