Hacker News Reader

Comment by amelius | Hacker News Reader

Yes, but that's not an innovation worth billions of dollars, like real high tech innovation that takes billions of dollars to get off the ground (e.g. the semiconductor industry back in the days). Instead it's pretty much straightforward functionality that could be implemented by any small team of programmers.

View on HN

> that's not an innovation worth billions of dollars

Small ideas at scale can easily be worth billions. Marginally-improved ball bearings, for instance. Or the idea of a supermarket.

Until they can sell their service profitably, whether it is truly “worth” billions is questionable.

That's the problem with today's economy. Investors are not using their dollars to support true innovation, but instead they try to grab markets.

I’ve felt that way about Twitter since the beginning. The core idea is like 1000 lines of RoR. Yet year after year, nothing displaces Twitter. Network effects are real and powerful.

Network effects are crucial to social networks like Twitter. But why should that apply to Uber? The network effect on Twitter etc. comes from the fact that each additional user increases value for existing users. But Uber is a two-sided market, and while each additional driver benefits existing riders they are also additional competition for existing drivers.

The article makes that point, too:

> [Uber etc.] also have none of the Facebook-type network effects (following what is known as “Metcalfe’s Law”), by which each new user makes the network more valuable to all other users, which makes it nearly impossible for smaller competitors to survive.

However, I think that's wrong. If I'm in a city with two "ride sharing" companies, but one has twice as many cars on the road, I'd expect to wait longer for a taxi from the company with fewer cars, right (or, if I check both, most of the time the one with more cars will show a shorter waiting time). Thus, as a customer I'd be inclined to choose the one with more cars.

But shouldn't most drivers be working for both companies?

I mean it was recently decided that uber drivers are contractors (not employees), so they should be free to do that.

Yep, but all else being equal you would also choose the one with fewer other customers. So in terms of the first-order effects, each additional user (as in, rider not driver) actually makes things worse for existing users! There's the secondary effect that each additional user encourages more drivers to sign up, but while this might compensate for the first-order effect it's not going to overcompensate.

For another perspective, you can look at things empirically. Has Uber used network effect to dominate the market and force out competitors? No; Lyft, traditional taxis, and foreign ride-sharing companies are still around. To the extent that Uber is winning, it's because it has a better product (for a definition of better that includes burning VC money to lower prices).

Contrast this with Facebook: the only major similar product in recent history is Google+, which died pretty quickly. All other successful social networks differentiate themselves in some way, because they have to.

Network effect for Uber is having the app installed. It is more comparable to a central market for a single, homogenous good.

Drivers don't really compete with each other. Maybe in some extreme future case, but its hard to imagine driver/rider ratio going out of control. So far more drivers just means lower ETAs which means more riders.

Twitter is a free product that has global network effects and they deliver a digital good which has near zero marginal costs.

Uber, on the other hand, delivers physical services to consumers that cost money and they do not have economies of scale or other innovations (e.g. self-driving vehicles) which would help them lower price.

> they do not have economies of scale or other innovations (e.g. self-driving vehicles)

Not yet. They're on it, having invested about $1B so far. https://www.bloomberg.com/news/articles/2019-04-11/uber-has-...

Not on Uber scale. You should watch some Uber presentations from developer conferences before writing comments like yours.

lots of "uber scale" is self-induced over-engineering. I'm pretty sure for any enterprise piece of software that exists there is a parallel uber solution that does the same thing.

Given that their basic product is a smartphone app that connects car drivers and customers the size of the entire operation is quite ridiculous.

You should keep in mind that innovation exists in the context of the cost that is associated with it. Building giant software systems alone isn't enough.

Scale is all about numbers. Solutions that work for small scale don't always translate to large scale. Plenty of information about Uber scale is publicly available on the blog and YouTube.

Uber's application can be almost perfectly sharded because the problem they are solving is mostly local. This simplifies the problem significantly.

Even with sharding the scale is significant. The QPS is massive because everything is real-time. Optimized matching and routing decisions based on many factors are instantly made on a large scale even locally.

What to do if data center in one region goes down? Should Uber not support users in that region because of sharding? Most probably that's not acceptable solution.

> Instead it's pretty much straightforward functionality that could be implemented by any small team of programmers.

This is some pretty extreme gatekeeping on innovation. You don’t have to invent brand new technology to innovate. Applying existing technology to solve a problem is innovative. Coming up with any new way of doing something is innovative, regardless of how small of an engineering team you need to implement it.

Reply on news.ycombinator.com