Desks and Chairs Don't Belong in the Metaverse
Desks and Chairs Don't Belong in the Metaverse
There are far more compelling use cases for VR than simply recreating physical spaces
Very few products I have experienced have given me a sense of ‘this is magic’. Among these are playing beat-sabre in VR, and the remote presence tool gather which simulates a virtual office for remote work. Although these software tools currently have a magical feel to them, they haven’t found mass adoption yet. I think a lot of people feel like VR and remote presence tools are perpetually ‘just over the horizon’.
Maybe we just need the right combination of hardware and an existing distribution engine, as argued in Stratechery. I think this argument is basically correct, but the use case is wrong. A lot of people seem to have this idea that the metaverse will be great because we can meet in virtual offices without the expense and hassle of traveling.
From my perspective, this is like saying, hey the internet is great because it’s like giving everyone a really cheap, really fast fax machine that can send pictures and videos. It’s not really getting how awesome this stuff could be.
How to Make the Metaverse Worth it
When adults go to work, them being in the same physical location is much less important than being in the same mental location.
Forget meeting in a virtual office with virtual tables and chairs. What I think you really want are meetings in synthesized spaces that represent objects of enormous complexity. These wildly complex objects exist in many workplaces, and many professionals spend lots of time trying to get on the same page with each other about the nature of some complex system. I think this is the right problem for the multiverse to solve. If it’s better than the current real-world approach, even for people who can meet in person, people will use it.
I want to meet stakeholders on other teams in a digital representation of the data pipelines I work on. “This feature you’re asking for requires us to split this portion of the pipeline in two” makes a lot more sense when I can gesture to a visual representation of an ETL job.
I want to do code reviews with another engineer while navigating through the abstract syntax tree of our codebase, and show that suggested re-implementations are visually simpler because we can talk about complexity of the programs’ branching structure next to a visual representation of that branching structure.
I want to see data flowing through my code, with a profiler hooked up to a geometry generation pipeline. I want to hear stack frames humming up and down like lines on a hi-fi stereo.
I want to see database queries in operation, as they churn through data, with chunks of work represented by visual objects. I want to watch the hotspotting happen as Justin Bieber gets into a tweet war with Elon Musk about bitcoin.
See this tangled mess of code over here? Right now, even if I’m sitting next to another engineer, we can’t have a meeting of the minds until get them to mentally represent the same abstract structures that I am thinking of.
Having code render these structures on a screen has long been doable, but the absence of the ability to cleanly and naturally navigate those structures, with someone else who is there, wasn’t.
The metaverse can simultaneously automate the rendering of these abstract spaces, and then allow two people to use their locations in those abstract spaces as proxies for the current focus of their attention.
One big advantage of the physical world over the internet is that where a person is, in physical space, acts like a decent proxy for what they might like to pay attention to. If I find you in a bookstore, I can guess you’re interested in talking about books right now. If I find you in a pet store, I can guess you’re interested in talking about pets right now.
Why can’t doctors have a virtual meeting in the middle of a patient’s CAT scans, to talk about the area in question, or go on a walking tour of a patient’s medical history?
Why can’t accounts have virtual meetings inside representations of a business’s cashflows?
Why can’t financiers have virtual meetings inside a space of various projections of expenses and costs?
Who let the DAGS Out
If you’re a software engineer at enough places, what you eventually realize is that every company’s tech stack consists of a single giant DAG of computation, which is never explicitly spelled out anywhere.
Services and databases talk to each other, and each specific link of communication may be spelled out in detail, but tracing the overall flow of information through a business currently involves a bunch of human beings sitting around simulating computers with their brains. Companies with big tech stacks are like social networks populated by databases and services, but nobody can see the full graph at once. It’s like every company lives in some post-babel wasteland with all of us siloed on sub teams using specialized languages, and the overall flow of data becoming totally infrangible beyond some horizon.
I think we are only just getting started with how computing can boost human productivity; the key to getting VR into work as a commonly used tool is probably not to re-imagine existing workflows with computers in them, but to use computers to interact in ways that were previously impossible.
If you like this idea, cut out the coupon on the bottom of this fax, put it, along with a self-addressed, stamped envelope, in a letter to APXHARD, PO Box 12345 Fakesville, OH,USA.
Bieber argues that bitcoin’s proof of work is net beneficial to the environment because it acts as a flexible buyer for new power sources. Elon musk argues that bitcoin’s proof of work is better for the environment because it incentives capture of methane offgassing. The whole thing is deeply civil.
If I find you at a grocery store, complaining about inflation, i can assume you are interested in talking about bitcoin :)
At Snapchat, on an advertising team, they asked all the new hires “What is your favorite ad?” The first time I heard this question, I heard “what is your favorite DAG” and was totally ready to answer.
A team at Google built a thing called ‘unified network model’ which wasn’t a model, and wasn’t unified. It was a language for describing network models in graphs. There were three different network models, all of which were incompatible with each other, expressed in a shared language that conflicted with the vocabularies used on individual teams. Different models would contain conflicting descriptions of the same items. You cannot imagine how many conversations got derailed by different teams using different meanings of the word ‘device.’