Parametric/Generative techniques and softwares for producing an "average" type design

Primary tabs

7 posts / 0 new
Last post
Andrea de Franco's picture
Joined: 7 Apr 2011 - 9:13am
Parametric/Generative techniques and softwares for producing an "average" type design

Hi, I'm collecting feedback and help for my thesis project.

The research part of the project is an analysis of the ongoing tendencies in design, such as generative design, conditional design, parametric design etc.
As a result of the research, I'd like to design and publish a complete typeface to be used as the official font for the Academy I'm studying at.

I'd like to work with all the other students, researching their taste in typography, asking them what fonts do they think are more suitable to be officially used by the Academy, and then producing an average font based on the opinions.

I've read about various parametric "average" typefaces, like the Averia:
and the Average typeface

But I want my average font not to be based on my sole taste or on every font installed on my font library. I want it to be the true recreation of the opinions and ideas of all the students from the Academy (even those who don't give a sheewt about type design, yep).
But in order to do such a thing I still need of to teach any typedesign application to make an "average" type, whatever the sources of the process.

Do you have advices/tips/etc.?

Thanks in advance for the reply, this stuff is quite important to me as it would be my first outbreak into 'true' type design.


Frode Bo Helland's picture
Joined: 26 Feb 2007 - 1:03pm

The average taste in type gets old very quickly.

Nick Shinn's picture
Joined: 8 Jul 2003 - 11:00am

I recently taught a course of type design to graphic design students.
The goal was to create the students' version of a contemporary generic sans, through consensus.

One thing I will say: the taste that is expressed through the process of design is emergent, and dependent on the way in which the designer(s) use their tools, no matter how automated the process may be.

The taste that may be appraised by choosing preferences amongst existing types that people have seen is somewhat different.
And the taste that may be appraised by choosing preferences amongst existing fonts that people have used is different again.

So you are setting yourself an extremely tall order if you expect a generated font to be able to recreate the opinions and ideas of the massive.


There is another problem with averaging.
Say you're considering serifs.
If 40% prefer serifs and 60% sans, the average will be a semi-serif, which nobody prefers.
And it wouldn't be at all representative.

darrel's picture
Joined: 4 Feb 2003 - 6:03pm

I like Averia as it's a nifty little mix of computer science mixed with art and the end result is rather interesting in its own way.

It is what it is and it's interesting and looks nice. But it's not a treatise based on deep scientific analysis of the opinions and science of type searching for some profound revelation.

I guess what I'm saying is that your intent may be noble, but it's really way out there. I'd step back a bit and first ask yourself what problem or question are you trying to solve by figuring out what your classmates thing about what defines 'average type'?

My advice would be to ground it a bit. The reason things like Averia worked is *because* of a defined (if not arbitrary) subset of data was used.

(Also, this feels a lot like "what happens when we design by committee". That question, sadly, has already been answered and tends to be answered daily in most corporate settings.)

Riccardo Sartori's picture
Joined: 13 Jul 2009 - 4:20am
Hrant H Papazian's picture
Joined: 3 May 2000 - 11:00am

Even though the results are difficult to use,
this is a superb enabler of thought in design.
To me much more interesting than merely
expression for its own sake.

But -in anything- beware extremes!


Andrea de Franco's picture
Joined: 7 Apr 2011 - 9:13am

Thanks for the feedback!

My intent is to produce both an average serif and an average sans, so the research result would be splitted into the two categories.
I like the results of the averia, as the serifs and the sans have morphed into a very particular kind of shape (much like a gross version of the Optima corrections), but I'd prefer a more 'ordinary' and 'familiar' look.

I know that actual representation of a typographic taste is almost utopic (and the wider the chosen subjects, the more utopic the representation gets), but that wasn't the real scope of the project.

What I'd really like to realise is to let all the students participate by their feedbacks and opinions on my project, but since most of them do not have a specific education in type design, they'll naturally refer to existing typefaces in order to express their views and opinions.

I'd like to create some sort of memetic and psychological link in between the resulting typeface and the final users of the typeface.

Intermission/further explanation: the more I study modern design practices, the more I realise that in a complex system like nowadays society, the designer is just an operator among operators, while the old creators-to-users system is collapsing fast. This was the first spark that brought my attention to the fact that, in order to create a typeface that really works for the users-operators (i.e. academy students), I should give the highest priority to their feedbacks and needs. As a vast number of those students do not really care about type design, the first consideration would be to abandon the project, as the final users do not need a specific typeface for them. But this year my academy is also finally establishing a visual identity of his own, after fifty years of basically lying in a 'identity limbo', and in my opinion a typeface is a very important part of the visual identity of an academy where a good number of students do study type design, whether they care or not.