Tools for Online Speech logo
by Jacob O'Bryant

I can't believe I'm about to build another dumb web app

Last month I wrote:

At the moment, my top priority is to get The Sample at least to the point where it can support myself. I can't spend all my time on tools for online speech if I have to start freelancing!

Long story short, I think I have in fact reached the point with The Sample where the only path forward is to pivot. The core problem is that long-term retention is very low, and I think the only way to fix it is to build an entirely new product.

One thing I've learned from 3.5 years of full-time entrepreneurship is that I don't really care about it that much. It is a means to an end for me. (See The trade-offs of being a startup founder and We need a career path for invention.) As such, I've been telling myself for a while that if I end up needing to do another pivot, then I will "pivot" to being a freelancer and try to do it part-time so that I can use my remaining time to work on my ideas as open-source projects, without business constraints.

And so, it is with great exasperation that I must announce... I think I'm actually going to build another product and keep being a full-time founder. 🤦‍♂️. For quite a few months I've been mulling over a product idea (a reader app, which I'll describe below) that I really want for myself, and it would be complementary with The Sample. My plan had been to work on it part-time after getting settled with freelancing. However, yesterday while I was thinking about some freelancing leads and what my plans should be, I was hit somewhat suddenly by a realization: the design for the reader app has evolved and simplified enough in my head that, honestly, I think it would work for a broad audience.

In addition to all that, my second daughter will be born in three months, which provides a nice deadline, and I still have strong stakeholder (i.e. marital) support for being a full-time founder. Thus, I've pretty much decided that I should build the reader app, have a handful of people try it out, and then figure out if it's worth working on full-time after the baby comes. The #1 goal will simply be to build an app that's worth using every day indefinitely, to build something that is likely to have high long-term retention. If I can do that, I'm confident I can grow and monetize it, as we've done to an extent with The Sample.

I'll still keep an eye out for consulting opportunitiesif the right thing comes along, I'll happily take the money and use it to fund my work on the reader app laterbut I'll keep the bar higher than I otherwise would have.

The reader app design

The basic structure will be like so:

  1. The app shows you a list of five links to articles.
  2. You click on one of the articles, which is then displayed inside the reader app.
  3. After you read the article, at the bottom of the screen there will be another list of five articles. Repeat from step 1.

Crucially, the articles come from newsletters you've already subscribed to and other content sources you connect to. Unlike The Sample, this isn't meant to help you discover new articles/writers: it's meant to help you actually read the stuff you've already discovered, without ever feeling overwhelmed.

When you sign up, you'll pick a username and get an email address which you can use to subscribe to newsletters. Something like, but with a different domain (I haven't picked one yet). After you've subscribed to some stuff, we'll use a relatively simple algorithm to pick your five links.

The algorithm will basically just try to figure out which newsletters/content sources you tend to click on first. Say you have unread emails from 20 different newsletters. At first, the algorithm will pick five newsletters at random and include a link to the most recent unread email from each one. But then perhaps the algorithm notices that whenever "Jane's Cheese Newsletter" is in your list of five links, you always click on it. Then in the future, whenever you have an unread email from Jane's Cheese Newsletter, it'll definitely get included in your list of links, and the remaining four links will be random.

(That's a simplified explanation; in practice it'll all be probabilistic. The technical term for this is a bandit algorithm. I built something similar into one The Sample's predecessors.)

The last essential point is that this won't just work with newsletters, it'll work with as many content sources as possible. If you connect to Twitter, we'll batch the tweets into "articles", so one of your links might be labeled "the past 20 tweets from Joe Schmoe." If you connect to Pocket or Instapaper, we'll mix in your unread bookmarks. Maybe you can even import ebooks, and then we chop them up into 10-minute sections. And so on. The idea is to have a single place to read all your stuff, with a nice algorithm to help you sort through it with ease and panache.

Bigger picture

In the unbundling social media example, I described the four types of services which I think should be developed separately: publishing, consumption/reading, discussion, and aggregation. (I'll probably stick with "reading" instead of "consumption" because someone pointed out that "consumption" might make people think of the disease, and reading is mainly what I'm personally interested in anyway.) The Sample has been an experiment to see if this unbundled ecosystem can be bootstrapped by building an aggregation service first. My conclusion is that it cannot. At least not by me.

Instead, I think a better strategy is to build apps for reading and publishing first, and then build aggregation services only when the need arises. Reading apps should be built to work with content no matter where or how it is published, and similarly publishing apps should be built to publish content anywhere/everywhere. I think we'll be able to go pretty far with that before new aggregation services become a high priority.

Thus my plan going forward is to put most of my energy into developing the reading app, while continuing to maintain The Sample. If the reading app turns out as well as I hope it will, I'll continue developing The Sample as an aggregation service, and I'll have it support the reader app.

For example, The Sample could provide recommendations for new users who don't have any subscriptions yet. I'd also like to turn The Sample into more of an ad network (I started an experiment in this area last March). The free version of the reading app could have ads from The Sample, and I might use The Sample to advertise the reading app in other newsletters. And anyone else who develops a reading app would be free to integrate it with The Sample in the same way.

And to come full circle, I have also previously started an open-source publishing app which already has a few contributors. At some point I'll start a discussion app since I've had plenty of ideas for that too. Then we'd have an implementation for each of the four types of services. We can then ensure they all integrate smoothly and provide an experience that is hopefully as seamless as a regular monolithic platform like Facebook, Twitter, or Substack; all while keeping the services cleanly separated. Then anyone can develop their own individual competing services and still participate in the ecosystem.

And honestly, that is my plan whether or not I'm able to do it as a startup. If the next three months don't go so great business-wise, maybe I'll release the reader app as an open-source project and then also build open-source aggregation and discussion services. Startups are nice when successful because you can pay a bunch of people to work on your ideasbut they're not the only way to coordinate a movement.

Published 11 Jul 2022

I write an occasional newsletter
about my work and ideas.

RSS feed · Archive

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.