The Internet of the late 90s/early 2000s had a number of logistical limitations even by then modern standards. Internet connections were slow and data storage was limited, so as much as Web 1.0 was innovative for the time, even the most creative pioneers had to work within strict confines. HTML created the basic blueprints to build websites upon and CSS allowed for basic stylizing. Eventually JavaScript added more possibilities to the infrastructure, but given loading times and storage sizes one still had to work within confines. If you’ve never seen an early web site, Wikipedia hasn’t changed much of its design in twenty years. Neocities is contemporary but is built to allow users to design in a Web 1.0 aesthetic. Doing a bit of digging, I found a website that at least claims to have been owned by the late Sean Connery and doesn’t seem to have been updated since 2004. I’ve come across Heaven’s Gate a few times on nostalgia/inspo boards, although I’m still not sure what it’s supposed to be. Coming Soon Magazine posted its most recent update on June 7 2000, although somebody is still managing the site’s multiple ads per page. The Old Net hosts emulators for earliest versions of now famous websites, although I’ve had varying success with trying the links.
The Old Net’s Amazon emulator shows an approximation of what one would have found using the online retailer during the early aughts. (Given the Motorola Razr V3 promoted for a rebate atop the page, this page dates at least past 2004.) Not much more could be done with the site than display the product selection as is. The site practically recreates a product catalog you’d find in print for a brick and mortar from the same time. Brick and mortar’s major advantage over online retail at the time can be clued into by a prompting above the Razr deal: “Sign in to get personalized recommendations”. Using a machine learning process called “collaborative filtering”, early-aughts Amazon’s personalization program was still in its infancy but would soon become important to their own, and all of online retail’s, success. Much as they try, pre-machine learning online retailers couldn’t match the personality experience of being greeted in-person at a brick and mortar. As consumers we intuitively understand that retailers have overcome this problem and then some, but it took time to work it out.
I recently patronized a newly opened local bookstore in my neighborhood. I’d been meaning to make the visit for a while--not many bookstores open in my area and far fewer open as independent retailers separate from megachains. Like Meg Ryan in You’ve Got Mail (1998) I tout myself a local hero every time I resist the Fortune 500. Self aggrandizing and having a bit of change to spend, I entered intending to make a purchase despite having not knowing their catalog or knowing what I wanted to buy.
The Google Maps listing I’d searched days before told me the store was Black owned and had a collection of secondhand works, but the secondhand part was wrong. The store sold books new in that they hadn’t been purchased before, but had a collection of genre old and new. On one wall children’s books all with Black main characters filled every shelf. Another filled with biographies and autobiographies from all of our prominent heroes- from Alex Haley’s The Autobiography of Malcolm X to Michelle Obama’s Becoming. Turning to the opposite corner I found all of the writers I’ve always told myself I’ll read but haven’t yet: Alice Walker, Zora Neale Hurston, Ralph Ellison, Langston Hughes...
After a few minutes of watching me peruse aimlessly the teller asked if she could help me find anything. She’d caught me off guard and I hadn’t mentally prepared for basic conversation so I lied and told her I couldn’t find the fiction section, but truly I didn’t know what I was looking for. Familiar with the store’s collection, she guided me to the wall clearly labeled fiction and talked me through the selection. Atop she pointed me to a book she’d finished recently, Octavia E Butler’s Kindred, where a Black woman in the 70s finds she has the power of time travel but only to slavery. She showed me the section of romance novels that I promptly dismissed and then to the newly released fantasy. She recommended Tomi Adeyemi’s Children of Blood and Bone, the first of the Legend of Orisha series that has since been acquired by Paramount for a film adaptation directed by Gina Prince-Bythewood. I purchased Kindred, although I did go back to Barnes and Noble for Children of Blood and Bone (I had a gift card).
Unlike the narrow curation strategy of the independent bookstore, large chain retailers take a blanket reproach to housing as many permutations of library as possible. The online store takes this further, hoarding a theoretically infinite library as warehouses and shipping make supply and demand effectively moot. All of the world's print ready at a click of a button with third party wholesalers organically integrated for anything not housed onsite, and it's available to all users at all hours of the day. Yet, it's not perfect.
Entering that bookstore I didn't have anything in particular in mind. I browsed with an unadulterated openness, but genuinely if I hadn't been directed by the clerk I never would have gone to the fiction section. My general tendencies would have led me towards the psychology or political science sections--I can’t allow myself to have fun. I never would have picked out Kindred on my own--I didn’t even know it existed. Online, an algorithm would have been even worse. With its personalization metrics intent upon showing me more of what I already like, another bearded white man would have been in my cart before I ever even saw anything else.
In building the original online retailers Amazon and others succeeded in promising customers their infinite supply, but the challenge came in finding a way of directing customers toward their next purchase. If you went on Amazon knowing exactly what you were looking for, great, but where is the random find that shocks you out of your immediacy? Curation became the question. You could point customers towards bestseller lists and editors picks, but one size fits all won't work for everyone. To get your customers out of their comfort zone enough to pique intrigue was the goal, but they needed a sweet spot of relatively similar but comfortably different. Methods of personalization became the solution. Algorithms were their mode d’emploi. But algorithms suck. At their best they can only use easily computable metrics to service their functions, which is terrible for works of art.
Algorithmic methods of cataloging became the default for online retailing. In attempting to simplify the process, every product had to become a category, neatly placed within a few preordained boxes. With the algorithm-first mode of thinking, everything became content. And content… well that’s an entirely different conundrum…