Discussion

By Joe Treacy on Jul 19, 2024

Agreed, Frank, that an even faster and more exacting type design classification system can be useful.

For most type users, time is at a premium all day, every day. And they don’t need a classification system that actually gets in their way.

Ultimately, unless one just wants to spend an afternoon leisurely browsing typefaces, it’s instantaneous culling and visual comparison that really matters to most time-starved font users.

Of course, AI is here to help sort and order analogs for faster, more accurate comparison.

To your point about current systems being a little too coarse, and for those who might not be aware of it, the coding world’s “css font attributes” system could use several more layers of descriptor variables to help AI move even faster, comparing fonts.

Today, that numerical “font-family” and “font-weight” system is probably the best system to build out from, since it’s a cornerstone of css web page coding relied on worldwide.

Even that circa 2003 system is a little too undefined, with (similar to what you mentioned about older current classification systems), Serif, Sans-serif, Monospace, Cursive, and Fantasy (rather than the traditional Decorative description.

The coding for this kind of recognition, sorting and presenting a semifinalist list of options is a lot like the facial recognition from camera-captured imagery we hear to much about.

The inherent problem is that a very real human condition called “paradox of choice” is already bad with one million fonts out there. Users suddenly stopped in their tracks the paradox of choice are slowed down and unable to make any selection.

And of course, that can cause many users to just fall back to “what they’ve always used”, rather thsn finding something new that could help their layout.

The real elephant in the room slowing down comparison are the hundreds of thousands of font lookalikes and knockoffs.

It’s the very number of knockoffs and lookalikes which have long been weighing down font catalogs of companies who chase quantity over quality, that also slows down people simply trying to make a semifinalist or finalist selection.

Of course, the specific problem with font knockoffs has always been that only one or two character shape areas might have been changed to avoid infringements. That in itself has created the need for AI to step in and help compare and contrast immediately.

Of course, a case could be made for adopting a classification system with fewer options. (Are so many traditional classifications even relevant or needed today’s composition environments?)

For example, in looking at how to speed up type selection nearly fifteen years ago, by 2010, I started coding AI-matching approaches into my https://treacyfaces.com Typesampler’s search functions, and have further increased its AI effectiveness, since.

To do that more efficiently than earlier systems, in our CAT (category) section, I actually use fewer categories: only: Sans, Serif, Fantasy, and Utility.

Using Utility allows for separating out pi and border fonts away from alphanumeric Fantasy (Decorative), and get users immediately to visual internal search results.

Joe Treacy
President & Director of Typography
Treacyfaces.com