On My Mind # 4
Technology is making us crave the similar, and the unique
5 things on my mind - by Michael Dempsey
I promise I’ll get the right cadence down on this thing at some point. This email spawns from this thread. The process for this will evolve but as you'll see, some thoughts are random, and most are unfiltered or poorly edited. Either way, let me know what you like, don't like, or want to talk about more.
5 Things
1) Technology is making us crave the similar, and the unique.
This turned from a thought into a longer form blog post which you can read in its entirety here.
—-
Over the past two weeks I've read multiple pieces that are on opposing sides of what we’re craving as consumers and in our digital self-expression. They all speak to a duality of how we are both starved for individuality as well as driven towards homogeny. These points are communicated to us through some interesting trends in technology, social media, and pop culture. Let me walk you through my sequence of consumption here.
Beauty_GAN (not to be confused with BeautyGAN) is a sparsely documented implementation of a GAN that utilizes instagram makeup trends and then generates new styles, which Dazed put on Kylie Jenner's face. The resulting imagery feels very cherry picked (not a rarity based off of my experience with GANs) but the key point of the article is that these dataset sizes and inputs are continually built with human in the loop biases (we'll talk about this again in #4). Put more artistically in the article:
"...Beauty_GAN is like a mirror of popular culture, but the reflection staring back at you might not be what you expected. We teach a machine to see us and what it shows us back is not always what we see ourselves.”
Despite what the GAN mirrors back in terms of how "dystopian" something looks, there are two other related pieces that speak to this homogenization of taking the internet's makeup kingmaker (Kylie) and re-painting the internet's makeup onto her. Or as Dazed wrote:
"One could argue that, of all the beauty imagery we see on Instagram today, Kylie Jenner’s face, her aesthetic, holds the most influence. Every time someone copies her contour or lip liner there’s a further proliferation that happens. She influences what we think of as beautiful, what exists on Instagram. The Beauty_GAN project sees this inputted into a machine, and then lets the machine take over; the machine creates what it thinks is beauty imagery, and then paints it back onto Kylie’s face. And so, the feedback loop closes."
And this feedback loop has proliferated into other forms of celebrities as Telegraph highlighted in this article.
"Lil Miquela, after all, is the ultimate embodiment of homogenised, Instagram-friendly beauty. An ambiguous mix of different ethnicities, with on-trend freckles and a body that can be shaped and moulded depending on the body parts required, she can be everything that consumers desire at any given time. "
But the key here isn’t what makes a character like Miquela or Imma.gram compelling. There’s time decaying first order interest points of “is this a robot or a human?!” and the general intrigue of a synthetic being, but then there’s the natural, more commonplace feeling amongst many influencers of “they are kind of like me, but better.”
And this close similarity is what I believe is akin to the dopamine rushes that gamification experts have hit on for a long time of being partially satiated, but not entirely fulfilled. This phenomenon has partially been described as Selfie Harm. It keeps us wanting more, liking more, swiping more, for something that we know we likely can’t obtain. But what happens when we can?
When we’re given tools that allow us to have those “on-trend freckles” of Miquela, or the contour and sizing of Kylie’s lips, or the dyed hair of Imma.gram, then what do we crave? Perhaps difference.
This is what I believe we’re seeing in pockets of the internet today. We’re seeing massive share numbers generated by very differentiated and unique AR filters that eschew traditional beauty trends. As one of the creators of these filters says in a Dazed profile:
“These filters can be used in creative new ways that partly break with the expectation of self-depiction on social media…Breaking fixed thought patterns on how we perceive gender and beauty is important and much needed.”
Maybe this is just how we cycle influence. We tire from the popular aesthetic/approach, early-movers push towards a new approach, some subset of the influencers make the jump, while new ones are borne, and on and on we go. Or maybe early pockets of culture are at a pivot point of individuality because for the first time we don’t get to escape and recharge our batteries from the influence.
Or as Oliver Sacks put it best:
“(We) have given up, to a great extent, the amenities and achievements of civilization: solitude and leisure, the sanction to be oneself, truly absorbed, whether in contemplating a work of art, a scientific theory, a sunset, or the face of one’s beloved.”
2) Is it worthless or a secret weapon to be elite at investing in non-technical teams?
I'd really like to learn more about investing in non-technical teams. I think this is a skillset that very few people have, and many probably argue is useless in 2019. I'd imagine there's a big market inefficiency of pricing these deals though if you can nail ability for non-technical founder to manage a technical team and/or hire that manager post-raise. Counter-argument would be, they should be able to inspire enough to get tech lead pre-raise but that's a privileged argument IMO.
3) Are Creative Technologists ideal early engineers?
In some of the areas surrounding computational creativity as well as consumer there's a heavy need for good graphics engineers, 3D artists, and what people are now deeming "creative technologists." It's been amazing to see this cohort of creators emerge with cool projects spanning AR/VR, 3D, and AI/ML, often with a beautiful portfolio of cool freelance projects. Many in this space are naturally drawn towards these people, but I'm increasingly bearish on long-term freelancers being early hires as I worry about pace of iteration and their ability to bang their head against a wall for a multi-month time horizon vs. jumping from interesting project/ tech proof of concepts every few weeks.
4) Who designs the model matters a lot for structuring industry-specific ML models.
This paper matching influencers and brands. The interesting part is it's trying to say how to match influencers that are most similar to brands. This feels like a fundamental misunderstanding in the sense that a lot of influencer marketing is amplification, but also expansion (especially within micro influencer categories). They address this a bit but saying that the algorithm could analyze category types in posts, but it shows that humans in the loop for some of these ML models are increasingly important to get good domain-specific results that can ever be used. In addition the dataset is so small. I'd love to see this expanded to significantly more than 20 accounts but focused on just 3-5 categories.
5) This was a long newsletter with just 4 thoughts.
A few papers/links this week
Bringing Cartoons to Life: Towards Improved Cartoon Face Detection and Recognition Systems - Being able to use face recognition to identify a given character feels like an inevitable future as characters live across multiple platforms and crossover events continue to happen within cinematic universes.
BeautyGAN - Style transfer for makeup. Pretty incredible results and also an implementation that could only further push us towards selfie harm.
If you have any thoughts or feedback feel free to DM me on twitter. All of my other writing can be found here.