I struggled to find a nice term for this blog (partially because I brought home a sinus infection from one of my recent vacation trips), but also because I didn’t want it to be overly insulting. Well, other than myself who knows that you have to say more than a short sentence that doesn’t so much attempt to scare people into doing something, but rather a long paragraph that tries to scare the audience in a way that is also convincing.

I drove home, I started to think about how I don’t have the tolerance to talk (or write) like the people I hear so often using 100 words where I might just use 10, or even sometimes just a grunt of displeasure.

Even in my really long books and blogs, I don’t spend very much time convincing people that the techniques I suggest they use make sense. More like I spend a ton of time explaining how to implement my suggestions.

So I decided to try a quick experiment

Something I say rather often is a terse statement like:

“How does everyone not realize we need to normalize our database designs so our data won’t be a complete mess.”

To the point, and I am more than happy to explain normalize to anyone who wants to hear about it for hours or 100s of pages. But to try to make this statement like someone trying to set strategy, I just don’t have the words. So I thought, how might a LLM reword this.

So I went to my free ChatGPT client and asked for some changes to my wording:

“Reword the following blunt statement as a lengthier statement a manager would state it as a strategy statement that explains what everything means”

And I pasted in my earlier statement. It took me two tries because the first response was a bit short, so I added lengthier. I didn’t want it to be ridiculous, but enough to some convincing and I wanted more explanation than my first try, This was the output:

Here’s a more executive, strategy-oriented version that explains the rationale rather than simply stating the conclusion:

As we continue to expand our systems and the volume of data we manage, it is increasingly important that we adopt a consistent, normalized approach to database design. Normalization is not simply a technical best practice—it is a foundational strategy for ensuring the integrity, consistency, and long-term usability of our data.

By designing databases to minimize redundancy and clearly define relationships between entities, we reduce the risk of conflicting or duplicate information, simplify maintenance, and improve the reliability of reporting and analytics. A normalized data model also enables applications to evolve more predictably, makes integrations less complex, and reduces the overall cost of supporting our systems over time.

Without a deliberate focus on normalized design principles, each new solution has the potential to introduce inconsistencies that accumulate into technical debt. As these inconsistencies grow, they become increasingly difficult and expensive to correct, impacting data quality, business decision-making, and our ability to scale.

Our strategy should therefore be to establish normalization as the default design standard for operational databases, with any intentional deviations being documented and justified based on specific performance or business requirements. By investing in sound data modeling up front, we create a sustainable foundation that supports current business needs while enabling future growth and innovation.

Conclusion

Somewhere in the middle is probably the right way for me to think, and I might even use a tool like this to get ideas if I have to go into “convince people who don’t understand that something is useful even though it costs more and the main thing you will ever notice is that it takes a bit longer to finish writing the software.

(I will save that for another editorial one day.)

Finally, does this seem like AI? Maybe some? I would need to know who the supposed author was. Uf you hear it coming from me could you be certain, it was, but I also felt the same way as I read this as I do a lot of tech marketing papers trying to convince me to look at a certain technology or pattern for many years.

But like most things, at a high level it is easy to get things right. As you get closer and closer to theory and practice, it is harder and harder to rely on databases like those used by all the LLMs that were generated from all the content on the internet.