Skip to content

Saying Goodbye to Winner-take-all Software

Today software is a winner-take-all industry dominated by best-of-class products:

  • GMail for e-mail
  • Slack or Discord for chat
  • Stripe for payments
  • MS Office or Google Docs for documents
  • ...

Generally we see 1-3 products dominate any particular problem space. This makes sense; software is hard to build and free to replicate, so as a product outpaces its competition it gets more resources and invests further (positive feedback) and then distributes that excellent solution all consumers. Everyone gets the same best option.

Other industries differ though; consider clothing/fashion for example. There are hundreds of options for essentially equivalent clothing products. The barrier to entry to make a competitively good pair of pants isn't that high relative to say, building GMail. A small family business can rival the largest incumbents.

This is good. We get lots of different options catering to different tastes.

Software is becoming more idiosyncratic

We can all build our own GMail now

AI makes software easy to build. While GMail took an army of engineers during Google's golden age, I'm pretty sure I could build it today in a couple weeks with Claude. Probably so could you.

In fact, we could each make our own GMail. I wonder how they would differ? How would your taste and mine result in different products and how would those products appeal to different audiences?

I'd like to live in a world where we had as much choice over e-mail as we had over which pair of pants to wear

There's a lot of potential value here. For example my e-mail practices differ greatly from my elderly parents, and yet we both use the same product.

  • What would e-mail look like if it catered exclusively to developers?
  • What would e-mail look like if it catered to the elderly?

Given how integral email is to all of our lives, differentiation here seems valuable but absent.

Funding will change

VCs don't fund clothing brands.

Consider Slack as a business. What is Slack's moat? There's an ecosystem of plugins and some network effects, true, but mostly it's that it would be annoying to rebuild Slack. That's no longer true. Heck, I might get excited and rebuild Slack this weekend just to prove a point.

  • Would I have invested in Slack a decade ago? Absolutely.
  • Would I invest in Slack today? Absolutely not.

Winner-take-all economies are high-risk/high-reward, which creates a strong partnership between software startups and venture capital. Some of that will surely persist, but I predict that this partnership weakens for the simple reason that the moat for most software products has been the technical difficulty of reimplementation, which no longer exists.

This change isn't bad. Software is no longer expensive, so we don't need millions of dollars of capital to build cool stuff. However, we also shouldn't expect to make trillion-dollar companies. Once a product does well it should expect similar companies to pop up with different tastes.

Today it is probably easier to build a $10M software business, but more difficult to build a $100B software business.

Starting a software company may start to look less like starting Microsoft/Facebook/Google, and may start to look more like starting a clothing brand or design studio. From my perspective that's great.

Open Source will change too

This probably applies to open source software as well.

One reason why Python had such a strong community culture was that building performant software was really hard in Python (it's slow), and so we were forced to organize around a few common libraries (numpy, pandas, cython) that gave us C-level performance for free. This forced organization was critical to the growth and health of the ecosystem. Other ecosystems where performance and abstraction were easy, like C++, didn't have nearly as much cohesion (at least not from my perspective), which stymied their growth.

While I encourage people to collaborate, I don't know that it will be important to share software in the future. I do think that it will be important to share ideas though. Blog posts outlining ideas in sufficient detail that agents can re-implement the idea for each user might be the future of collaboration.

Final thoughts

Making predictions these days is hard; probably everything above will be proven wrong next month. But if what's written above is true, then the main lesson I think is that we need to moderate our expectations:

  • Old: First movers have an advantage and a strong possibility to dominate the market, generating fame, glory, and wealth

  • New: There's no rush. Everything is both easier and less important than it used to be. You don't need to reinvent the world today, but instead think more about what you and others would most enjoy.

Comments