Where have we come from in the development of the internet, and where might AI be leading us next? I don’t know, but I have some thoughts.

Early expertise created a baseline for quality

When people first began to venture onto the nascent internet, there was relatively little content. The earliest users were professionals working in scientific research, and governments. As the floodgates were opened to the masses, the tools, and platforms available to publish content still required a certain expertise. That created a limiter on who was publishing content.

The difficulty of finding content pre search engines, and really pre-Google, because the precursors were pretty awful, also meant the occasional crank spewing their manifesto onto the internet wasn’t being encountered by random people who might get drawn in.

I can be a very verbose writer, long run on sentences meandering across lines, with no sign of when or where the thread of an idea will finally terminate its arduous journey to a conclusion. Did that make you tired? A few years into my career, my manager pointed this out to me.

I had reached a point where I was communicating daily with senior and executive management. I’ve also worked almost entirely in startups and smaller engineering departments. Everyone is busy all the time. Presenting information in the most digestible format is critical to get your point across. My manager, in a brutally honest piece of critique, told me he rarely reads beyond the first line of anything I wrote. His advice, focus on short declarative sentences.

For the last 18 months I’ve been working in an environment that uses Github Actions as its CI platform. I don’t have good things to say about it. Outages aside, of which there have been quite a few, Github is developing Actions the way many companies would develop a consumer app.

There’s a regular cadence of breaking changes. Some come with warnings and deprecation notices, like the change to how outputs from steps are handled. Others, our team wakes up in the morning and finds out an update has made our version of Python unavailable.

It feels like we’re sinking way too much time into keeping up with changes forced upon us for a part of the system that desperately needs to be set and forget 99.9% of the time. The collective productivity lost across the industry to Github’s CI changes probably adds up to billions of dollars a year.

I wrote this a few years ago to share with friends and colleagues. It got a lot of positive feedback. I figure I might as well post it here.

Why do we review?

There are many reasons to request and offer code review. It helps coalesce our many variations into a gestalt style, building our culture and interpersonal relationships. It results in higher quality code that benefits the company and the code’s future maintainers. It distributes knowledge to increase our organization’s resilience to personnel changes. On a personal level, it grows our own abilities; this is true for both reviewing and being reviewed.

If you’re outside the USA and don’t know what “Section 230” is, in short, it’s the bit of legislation that means social networks (and other online publishers/platforms) aren’t accountable for the content their users post.

It has some limitations that everyone can easily agree on. If somebody posts illegal pornography for example, the platforms have to remove it. There are however still a great many things that fall into a grey area based on subjective opinions. It’s commonplace to hear people from all over the political spectrum calling for somebody to be banned from twitter for something they wrote.

While we can all likely point out some people we would rather didn’t have a voice and platform, it’s nigh impossible to define, at least in any way sufficiently long lived that we would want to codify it in legislation, what is sufficiently controversial to warrant censure.