Asleep at the Wheel

27th October, 2026

One of the reasons why things like WordCraft are such an indictment of established software—in this case Microsoft Word—is that some of the supposed best software development teams on the planet have been caught napping.

I use Word as an example of this because I'm really familiar with it. Word has not been capable of handling my 1000+ page books for quite some time now. Why? It never made it past 32-bit data processing (currently, all OSes and virtually all apps can handle 64-bit files).

This is one of the clearest indictments of Microsoft's Office team that can be made. Yes, rewriting code from 32-bit to 64-bit is a big chore that doesn't seem to generate any new users or appeal to 95% of your existing customers. So you get lazy and don't assign any resources to that.

It's akin to what has happened with some of the road structures here in the US. Yes, rebuilding the highway system and bridges doesn't seem to do anything except slow traffic around construction zones, but in reality, infrastructure is terribly important to keep current and functional. 64-bit would be a software infrastructure investment Microsoft never made, yet we're all using machines that can handle that. Indeed, prefer that.

I can give another personal example of a similar mistake. PageMaker once dominated the publishing scene, both personal and professional. A company I started had designed what—even today—was a full-fledged and remarkable product that could do tables better than anything else available. One reason for that was we spent most of our time on infrastructure that the user never saw. Tycho TableMaker could do things that I still don't see modern table programs try to tackle.

The maker of PageMaker, Aldus, wanted to buy Tycho. It would have been more than a million dollar payday for me, and would have paid back our few angel investors very handily.

We got all the way to due diligence, at which point both companies put their code on the table to inspect. Our first question was this: "we don't see how we can hook into your text handler, how do we do that?" Tycho was text agnostic. We could plug any text handler into it. Again, infrastructure anticipation. Text handlers change with time.

The answer stunned us, and basically curtailed the deal. That answer was essentially "the guy that wrote that code left, we don't know how it works, and every time we try to touch it it breaks, so we don't touch it."

You may recall that Adobe bought Aldus and PageMaker. I don't think they really did their due diligence before making that acquisition, because Adobe already had Framemaker (which has an excellent text handler). I suspect that they saw that they could conform the long-form abilities of Framemaker with the short-form abilities of PageMaker. But they couldn't, which led to Adobe coming out with yet another page layout program in InDesign, and eventually terminating PageMaker.

So the real question about whether packaged software will sustain through the AI vibe coding that's going on is simple: have the software companies been sleeping at the wheel assuming what they've got is fine and doesn't need infrastructure investment, or have they been building new architectures and scaffolding to support current and future technologies?

That last part is important, because AI coding is looking backwards, not forward. AI has been trained, and trained quite well in my experience, at dealing with all the functions and algorithms that have appeared in the past, particularly in Open Source programs, but doesn't do so well at anticipating how new tech will work in the future.

I don't count a lot of software companies that are pioneering any more. Adobe's one of the better examples that does. Moving to subscriptions was one of the things that's allowed them to do so, as it brought a reliable stream of revenue that wasn't subject to "good update/bad update" cycles. However, moving to subscription is also a long-term problem, as AI-coded things start appearing that do 50%, 60%, 70% of the user needs, those subscribers will stop subscribing.

Thus, it's tough to predict the future of packaged software. If pressed, I'd say this: you need to be fast, nimble, creative, and always moving forward. If you can do that, then, yes, you can stay ahead of the coming AI-software onslaught.

But a big juggernaut like Microsoft, which has squandered their dominance in "office" software? Sorry, but they're too slow, not at all nimble, uncreative, and usually moving backward (yes, features have been removed from Word).