Focus in the Shell

Recent technological advancements have introduced a wide variety of specialized software tools that supposedly enhance productivity and help to overcome the age of paper. However, in essence, most B2B software as a service companies offer tools that are not associated with any additional value. The terminal on any UNIX-based system offers most of the functionality an average person could wish for, while it is free of charge and distraction. The simple beauty of the terminal allows the mind to focus on those matters that are of true importance and that do not require a designer obsessed with capturing attention to make content worth reading.

We have entered a dystopian world in the fight for the attention of consumers. It is not the diabolical plan of an evil Machiavellist that controls the brains of most people; the sales teams of big tech have made humans’ most valuable possession their product. Their attention and time.

This is no new information. We know the algorithms of social media and the tracking systems of the world. And there is an entire business sector concerned with attention and focus-improving technology. However, and this argument is difficult to sell, as no one would ever earn any money with it, the liberation of most attention and software cost consumption may be attained by opening the terminal and working from there.

While tools for editing code in a graphical user interface simplify the entrance to the development world, they quickly come up with serious limitations. They are a product of design; they grab attention instead of funneling it into your project. They need updates, plugins, and in some cases, even cost money. The peak of absurdity in this regard has been the advent of AI integration in IDEs. LLMs in IDEs are proven to cause severe damage. They are convenient and promise quick solutions. You may be able to deploy a web app within minutes with the help of Claude, GPT, Gemini, or any other model that is sold to you, subsidized by venture capital. While the large language models consume vast amounts of energy and literally burn money, they will break your infrastructure if you let them. When I allowed Cursor to execute code in my terminal, it broke servers, changed sensitive configuration files, and sometimes introduced changes that I only saw when I looked at the git diff file. Cursor is a curse.

Writing one’s own code means one needs to go one step back in order to move three steps forward. The usage of LLMs in software engineering is undoubtedly significant. But integrating LLM-generated code promptly in the codebase is like having a chat about penny stocks with a friend in a bar and immediately buying anything you were discussing. Depending on the infrastructure one works on, the risk is comparable.

Since so-called productivity apps have started to integrate AI systems at the core level, the risks of this development have increased dramatically. I once loved Notion and used it as a life operating system. Now it has mutated to a system in which an LLM has access to all information on the platform, and the user experience has gotten so complex that it does not serve as a note-taking system anymore. If you need some kind of ERP without being willing to write code (or you have someone who implements the Notion API activities for you), it might be a viable option. But what functionality does Notion provide that neither VimWiki nor a self-hosted Odoo platform can give to you?

It is understandable that the product managers of the B2B software as a service companies do not care about your budget or your attention. They live on it. They thrive if you believe in their productivity evangelism, which measures productivity in Jira tickets. But your fear of working too slowly is their leverage to sell proprietary licenses to you. If you have any technical skill to escape from this prison of illusion, choose it.

The mechanisms of stealing attention and time apply to other fields in the software world, too. While I believed for a long time that YouTube was my teacher, I took my many wasted hours to understand that it is not. YouTube is an infomercial, and in most cases, it is hard to notice. I am not talking about beauty channels or technical customer advice on new products. Even if you watch content that seems to be seriously focused on content such as coding, hardware architecture, or even philosophy, the real home shopping show begins after the 20 seconds of ads at the beginning. I paid for a YouTube Premium account as I thought this would let me pass the ads, but in reality, the actual sales show started when I skipped the meaningless commercials at the beginning.

In coding, almost all channels only explain technology to introduce you to some kind of software tool. They are not educational; they limit you to depend on commercial tools that are only free for as long as you do not use them professionally. The best source on any serious subject is books, software docs, and some blogs with no affiliate links. There are exceptions to that rule, but probably fewer than you think, as the great salespeople never reveal their true intentions while they convince you of a need you never had.

As much as sharing means of production has failed historically (for whatever reason, you may believe this is the case), in the software world, a special set of rules seems to apply. As sharing code does not limit your personal resources (other than when you share a factory) but even increases its potential, as people will collaborate to contribute to its functionality, many historical breaking points of Marxist theory do not apply. While the subject of overlaps between Marxist perspectives on shared means of production and the Open Source Software movement is certainly interesting for detailed analysis, we will not dive into this subject here. However, we may draw the conclusion that most cases of software that is offered could be left for Open Source alternatives easily, but there is no marketing budget for advocating this idea. It is not a conspiracy why most companies are deep in vendor lock, and most people have a total lack of time while they have as many hours a day as any other person does - it is simple attention economics, the underdog has no advocates in the constant stream of marketing.

Those who happen to work in science or software engineering may seriously consider what they use their computers for. There are exceptions for field-related software, but most likely your productivity boost is not to be attained by deploying (and paying for yet another tool, but by opening your terminal.