First part here with context and rules of engagement. You know how this works. On with the show!

1. Metrics and models around social media amplification and recommendation:

I predict we will have agreed upon metrics and models to reason about amplification and recommendation algorithms similar to how we talk of GDP and productivity growth to talk of the “health” of a country.1

The last few years have seen a sea change in our understanding of social media. We are now aware of the challenges around misinformation, polarization, content moderation, filter bubbles, etc. However, it’s impossible to have measured conversations around these with any discussion devolves into opposing narratives.

I believe we need two distinct kinds of developments.

First, a set of agreed upon metrics. These should be simple, intuitive and quick to get directional stat-sig signal on. Having an agreed upon set of metrics similar to how we use GDP or productivity growth for the economy will help with more evidence-based, experiment-based and quantifiable discussions. This might need to pull from research on epidemic/contagion models, study of complex/chaotic systems and AI explanability.

Second, we need better models to help visualize, interpret and debug relevance and amplification. It is really hard to make sense of how information flows in a community and how algorithms can cause changes. As a ML engineer, you’re often limited to limited telemetry. I’m a big fan of Bret Victor’s work and believe we need a “seeing space” for recommendation algorithms that help product managers/engineers experiment with and visualize impact of changes.

Getting here will not be easy. Tech companies internally often don’t have the right instrumentation and we live in a time of strained relations between large tech, press, government and policy stakeholders. However I believe this is crucial for making progress on some of these hard problems.

I’ve been reading a lot on this topic recently (especially around epidemic models, chaotic systems, AI explanability) and if you’re interested in these topics, do drop me a note at


I predict we will see a new kind of media organization/platform that focuses on being objective over all other factors (editorial/speed to report/etc).

If you’re like me, whenever you see anything remotely ‘political’ in the news these days, you’re instantly running a calculus in your head. Who wrote this? What’s the agenda of this news organization? What’s a counter argument? What are verifiable facts? What’s the strength of the sourcing? What’s just an attention grabbing headline vs the real heart of a story? Every news story requires Googling around and finding alternate points of view.

I predict the rise of a different kind of “fact” or “truth service” (I do think these will be different from traditional news and journalism organizations). These will prioritize objectivity and accuracy over all else and rest their reputations/brand equity on this accuracy. The closest analogy I can think of is some combination of Wikipedia and long exhaustively researched investigative journalism with every mistake triggering something akin to a AWS-outage postmortem.

There are many, many questions here. How do you fund this and incentivize contribution? Does this have a business model that can work? Who is the customer that wants this at scale? How is this different from many media startups today (like Axios or TheInformation)?

All great questions to explore in 2020!

3.Hyper personalized financial products and advice

I predict financial products (and advice!) precisely tailored to your individual financial situation and lifestyle.

In my 13 years in the US, my life has changed dramatically in life events, family (now with one 11 month old in the mix!) and economically. I’m not the same kid in his early 20s from India with no US bank account I was when I first showed up here. However. I’ve always run into the same problem when searching for financial products and advice - my bank and my credit cards still think I was the same person I was over a decade ago. Especially frustrating when they have so much data on me!

I predict this will change rapidly. Here are two themes with sample ideas.

First, in advice from people with similar lifestyles and net worth. Imagine if you were always able to seek advice from a peer community of similar net worth, income and lifestyle.

Second, in products offered. My travel-oriented credit card from two years ago when I was doing over 100 flights a year is no longer great for me while if one gave me a discount on buying baby products, I would be the first one to apply. Imagine a card that changed reward schemes based on your life changes.

There seems to be a lot of room for innovation here.


This is an assorted list of things I don’t have expertise in but still believe in (hah!).

  1. “Consumer RPA”: I’m a firm believer in voice becoming a more ubiquitous UX medium. The obvious next question is to think about the services and actions regular consumers will be able to perform via voice. One observation a friend had was that most of today’s useful services are locked away behind web and app interfaces that aren’t very flexible (think your favorite airline booking system). What would it take for these to be exposed via voice? The shortest path might be to stitch these using scraping/scripting similar to how RPA in the enterprise enables existing systems to be “programmatic”.

  2. AI nationalism: I believe we will see AI being treated the same way we think of nuclear weapons or any other strategic national resource. Ian Hogarth’s essay on this.

  3. Material+process+system design: As software has taken over the world, we have fallen out of love with innovation in the material world (when was the last time we got excited over something the way previous generations got excited over plastic?) as well as the processes and systems engineering that go with it. I predict this changes. Dan Wang’s essay is my favorite on the topic.