Code Was Never the Problem

This should be a renaissance era for software. The arrival of agentic coding using LLMs has made building it easier, faster, and less expensive. A plain language description is often enough to allow an LLM to reproduce and build a fix for a bug. The ability of coding agents to see, measure, and instrument behavior and performance means people need not give feedback when attempting to replicate a feature or functionality. We have opened a portal to a brighter future, so why have we not stepped through it?

Facebook beat MySpace one reason was the constraints put on the design of user pages. At MySpace you could customize your page to have a multitude of backgrounds, colors, and layouts. Facebook offered you one. As a user this was a much better experience. The amount of mental work required to visit each page was less and granted you the freedom to focus on the content and interactions. This lesson has largely been ignored by large software companies that use Electron to impose their own design and user experience patterns on users instead of using the native patterns made available by operating system makers. Considerations like performance, memory use, and battery life take a back seat to the aesthetic tastes of the lead designers. Cities often employ planning boards and zoning codes to make builders respect the history and architecture of the place they are building in. You would have expected say Apple’s App Store to weigh in on this manner, but you would be mistaken.

Good design is of course secondary to business goals, so software companies have deployed the dickover. Instead of making it as easy as possible to do what you want, they are going to push you to do what they want. Any glimmer of positive user experience is easily sacrificed on the altar of this quarter’s KPIs. If you think paying for software or a subscription service aligns the user and the designer, you would be wrong. There is always a marginal dollar to be won or newsletter to push.

Security is the opposite of user experience and it seems the first shibboleth of security is to make users log in as often as possible. Instead of sending users PDFs of their credit card statements, notices from healthcare providers, etc. to their email we instead make them log in to “secure” portals to get this information. In an act of intellectual gymnastics that would impress Simone Biles we treat email as insecure for the purpose of receiving information, but secure for the purpose of sending the keys to read the information. Making folks re-enter their password every day feels more secure than every month or year but the reality is they have already likely had to enter their phone password before opening your app or website. The fact I need to go to MyChart to read messages did not stop UnitedHealthcare from getting hacked, but I’m sure it did a great job of getting folks to visit MyChart more.

Signing in is now hell. If I go to my bank page Capital One 360 I click “sign in with a passkey” which then opens up a dialog asking me from where I should provide the passkey. Instead of letting me pick, I have to choose “more options” if I have not stored it in Apple Passwords (though I have). Clicking continue then asks me to enter my computer password. So I enter that and click enter. Instead of one box with a username and passwords, Apple’s “user experience” for a Passkey involves three mouse clicks, two screens, and a password entry. No wonder social media is full of passkey backlash.

Over 15 years ago Apple released an ad criticizing Windows Vista for all its annoying security dialogs. I see Steve Jobs in that ad. From his perch in the corner office fighting for a better product and experience. Since his passing Silicon Valley shifted from having opinions to gathering them. Specialization has lead to organizations that do not interrogate the asks of peer stakeholders. Cheaper code should have freed us to build better software. Sadly this has not materialized. The bottleneck was never code nor engineers. Instead the abysmal attention span and taste of todays founders and coders leads to software that is half-assed instead of Mac-assed.