Monday, August 1, 2022

Strava Adventures

Currently, Strava's feed is composed of activities. Increasingly, however, athletes participate in multi-activity adventures - think bikepacking trips, hiking journeys and ski touring adventures.

Adventures pollute the activity feed by over-posting, and reduce the amount of social behavior by partitioning comments and discussion across semantically related activities.

Strava should introduce a new concept (an adventure) which groups together contiguous activities for the purpose of increasing the prominence and engagement on the feed while also helping users remember particularly meaningful adventures by selecting key photos to highlight the experience.



Friday, July 15, 2022

Excel-based Azure Functions

The world runs on excel, and increasingly, Excel documents are store in O365 / Sharepoint / OneDrive - in the cloud.

Microsoft should support turning any Excel document into the source code for an Azure Function so that you could have lambda functions for every single macro in Excel, easily available for developers to hit.

This would continue to drive people into Excel while also supporting the transition of businesses towards interconnected API-based companies.

Sunday, March 20, 2022

The pathology of Google's carbon-based flight metric

“A strange game. The only winning move is not to play.”

Google recently added a number of fairly odd features to their flight booking system that claims to help travelers learn how much carbon a given flight will release into the atmosphere (link).


But the numbers are misleading.

It's not exactly clear how they are computed, but even if they were correct on average, the irony of the feature is that a set of uncoordinated but climate-conscious consumers could end up increasing the aggregate amount of carbon for a given set of flights if they used these numbers as part of their decision-making process.

In fact - it's almost certainly true that price is a better predictor of the actual carbon emissions of a given route than the made-up Google metric. The cost of a flight is composed of a bunch of fixed costs which the airlines then try to make up on the margin by selling every seat on the plane. A full plane, broadly speaking, is the most carbon-friendly plane: and it's also the cheapest plane for everyone onboard.

Carbon emissions are obviously always lower for a direct flight assuming a full plane, but more direct flights between the set of airports would result in planes which are, on average, less full. There's some bin packing math to do here, but the basic premise is that while the naive calculation for me may imply that the direct flight from London to Austin only costs 731 kg of carbon dioxide, occupying one seat on both of the London to NYC and NYC to Austin flight may end up with fuller flights, thus reducing the per-capita cost of travel: both in carbon and dollar terms.

If consumers were to use this tool, therefore, the result could actually be to push airlines away from more efficient hub-spoke routes towards direct flights with lower utilization.

Of course, it's hard for me to actually believes that anyone actually think this matters: once you've chosen to fly to London, you've basically made the call that you're going to use a tremendous amount of carbon dioxide no matter what route you pick. And implying that the choice of flying direct is 20% better for the environment is just lazy - if you can book the flight, the planes are already taking off with or without you.

Climate change is not a problem that will be fixed by individual action: if we want to fix this problem, we need to pursue unpopular policies to increase the cost of carbon. As a consumer, trying to do multi-variate optimization for every choice is an exhausting way to make decisions; it's why having the almighty dollar, rather than a cryptocoin for every commodity, is the way we evolved out of a primitive economy into one based on the free exchange of goods and services.

Which leads me to the question: why did anyone at Google green-light this project?

The most cynical answer: it lets rich people (like Google employees) who can afford direct flights feel good about their ability to pay more money for convenience by giving them a hedonic boost when they book "carbon-friendly" direct flights (which again, they were going to do anyway, because they are actually paying for convenience). 

Sunday, January 9, 2022

When users are speculators: Can web3 apps generate sustainable usage?

Two of my previous posts were about the relationship between labor and capital in innovation-rich parts of the economy. Roberto recently asked: "I'm curious if you have any thoughts how the incentive structures change with web3 / crypto?"

I'm not a crypto bull, though I'm interested in the space, and I honestly think that crypto-economics has a related, but distinct modality of disruption.

In Observer or Participant I argued that investors, by betting on success, can reduce its likelihood. With crypto, I suspect that the same perversion can occur, at least in network-driven applications.

Consider a model in which users of a decentralized application, who have been granted tokens which increase in value as the application becomes more valuable. For the marginal user, is it better to a) participate in an existing application, or b) become a pioneer in a new clone of that application.

Without the user-owned tokens, the answer is clear - the value of the network will roughly track the number of users, with quadratic value a la Metcalfe's Law as an upper bound. The marginal user will receive more value from Application A than from Application B.

But in a crypto-economic world, the user's value equation is different: users receive value both from their usage of the application and from their token. I'm sure there's some math that we can do, but assuming a non-inflationary token (e.g. the first N users get the first N tokens), it's not clear that the same incentives hold. In some sense, you've turned your users into employees, and just like the investors in the previous post, these employees may find it more valuable to cash in and leave rather than staying - selling Token A at the peak and then joining Application B seems like it would be value-maximizing in many cases. And certainly for the marginal user, it seems rational to take a bet on Application B.

In other words - while it may be frustrating to users that Facebook's founder and employees have been able to capture so much of the value that Facebook's users have created, the new incentives in a token-based world may actually make it impossible to create networks with as much value as their web2 competitors; rather than winner-take-all, web3 network effects may actually drive a tremendous amount of internecine warfare with splintering communities forking and re-tokenizing each other's networks. Empirical results from the currencies (BTC / ETH etc) seem to imply at least some amount of stability, but we're in the early days, and the impact may not be what we expect - jealousy is a powerful motivator.