Laws of Tech: Commoditize Your Complement (2018) | Hacker News Reader

Laws of Tech: Commoditize Your Complement (2018)

gwern.net

156 pointstosh35 comments

I've seen descriptions of this strategy before, but this particular article lists way more concrete examples than I've seen before, which makes me wonder if it is now the default across the entire tech industry at all times.

Also: the strategy seems to depend upon a "greater fool," or just high risk-takers to fill in the complement with what is ultimately low or zero profit product. How can we guard against being that high-risk fool?

I mean, from the article, IBM tried to commoditize all of the peripheral makers by publishing standard interface documents. But then they ultimately lost out when Microsoft commoditized the entire PC by licensing their OS to other PC manufacturers. IBM didn't quite see this coming. They were out commoditized.

OTOH, you look at the way Microsoft was paying app developers cold hard cash out of pocket for listing apps in the MS mobile app store (which ultimately failed) and you see that MS was out-commoditized by Google/Android/Play Store.

So how can anyone be smart enough to spot the loser(s) in advance?

> I've seen descriptions of this strategy before, but this particular article lists way more concrete examples than I've seen before, which makes me wonder if it is now the default across the entire tech industry at all times.

If I missed any, I'm always happy to add more. (Sometimes the only way to really nail down an abstract pattern for people is to provide an enormous list of examples.)

well done. thank you. after i read it, i started to wonder if there are an equal or greater number of examples where companies tried this strategy but completely failed.

I recently finished S1 of Netflix's Narcos: Mexico, after reading your post the organization at the heart of the story comes to mind. A drug dealer who works for a single organization among many in Mexico unites (commoditize) the other organizations to distribute marijuana across the U.S border.

The distribution network becomes so large that when the organization branches out to distribute a new product (cocaine) and tries to set up a deal w/ one of two large manufacturers in Colombia the other one immediately lobbies for a distribution deal as well. Without the services of Mexican distribution the other Colombian organization would've gone bankrupt.

> Also: the strategy seems to depend upon a "greater fool," or just high risk-takers to fill in the complement with what is ultimately low or zero profit product.

It's the other way around; the commoditizing actor is that "greater fool", and others can then join in commoditizing the market out of pure self interest as consumers of the good. The agent which is commoditizing its complements is most important as a first-mover and solver of coordination problems.

trying to apply this, but not quite seeing how it fits.

if i'm an indie developer who thinks he can make money by selling apps on the Google Play Store, i put in a ton of work and then see little or no financial reward. i've made Google's Android platform a bit more attractive (because it now lists my incredibly beautiful tip calculator for free to all Android users). who is the greatest fool here?

> who is the greatest fool here? Not sure if this is a rhetorical question or not; it does sound like one.

In the case of a single indie developer making an app for a single app store, it seems the developer is the greater fool, since they are turning their work into a commodity, when they should be turning their complement --- the app store --- into one.

Perhaps the route for the indie developer to actually see the financial reward is to make sure their app is available on every app store. ... or maybe they need to skip the app stores entirely to keep 100% of the sales.

it's me. i'm the greater fool. (that's always the correct answer, btw. somehow it always turns out to be me.)

i wouldn't be so sure. it might be me. we might need to have some sort of "fool-off" to determine the greatest fool.

I don't really see how Microsoft was out commoditized on the phone. It seems like the causes for Microsoft's loss were more straightforward, it was late to the party, it was hampered by an existing product that wasn't great, and they mismanaged the hell out of the app store.

> it was late to the party

But Windows Mobile was on phones well before Google got ahold of Android and got the Google Play Store up and running:

In Q1 2003, Windows Mobile was the third largest operating system in the smart handheld market, behind Symbian and Palm OS.

https://en.wikipedia.org/wiki/Windows_Mobile#Market_share

The earliest version of the Google Play Store was made available in 2008.

https://en.wikipedia.org/wiki/Google_Play#History

Late to the party with a good smart phone. I used windows mobile and it didn't compare to Windows Phone or Android.

Iirc, they were charging for their mobile OS while Android was given away (only requiring the gsuite). MS was also later to launch an app store.

So the mobile OS layer is where MS use to make money and it's what Android commoditized.

Are you arguing that MS lost the smart phone race because they were too expensive? I don't know anyone who didn't buy a Windows Phone because of cost.

They lost because finding good Windows Phone apps was difficult. This is why I stopped using a Windows Phone, and why everyone I know stopped using a Windows Phone.

I could pick up a performant Windows Phone for $40, and I wasn't happy with my Android until I spent $200+.

Not to mention one of the largest companies in the world right now makes most of it's money off of selling a mobile OS's.

I loved Windows Phone 8.1

Handset was cheap, performant, the OS was gorgeous and easy to use, typing was great.

It's weird though. The apps are totally what did it, but I've found that I really don't use many apps on phones. My phone is littered with the, but my usage statistics say that I'm a slack ass when it comes to them. Basically just language applications.

Indeed, Microsoft already had the monopoly on desktop PC's when Linux took off so they could continue to hold it.

Trying to compete in a new market against free was a much harder prospect (and they where not the only one, I had a Blackberry Playbook, it was for it's time an excellent device and the OS was intuitive and solid...it got murdered in the market).

In fact unless you invent the category (iphone) you would be hard pressed to own it via the OS anymore.

Ultimately, it's commodities all the way down anyway. The only real "moat" that Apple's OS, MS's Windows desktop, Google services, etc. have is first-mover advantage, and that's not going to last very long.

Another way that I like to express that is "create a desert of profitability around you". I once had a strategy professor define the Google business model somewhat like that, where "Google tries to make every other business around it free or irrelevant"…A desert of profitability shifts consumers to you, and keeps competitors away.

Google seems to be doing this with YouTube and 10 minute online video commentary.

It should be called Spolsky's law :)

Yeah, he deserves this one.

Somewhat unrelated, but doesn't this principle apply very often in agriculture? I can see this as one of the reasons why only 12 plant species provide 75% of world food. [0] I should say I take it as evident that raw food is only a fungible commodity insofar as it adapts to market needs (which are not necessarily immediate consumer needs).

[0] http://www.fao.org/docrep/007/y5609e/y5609e02.htm

Another angle: https://stratechery.com/2015/aggregation-theory/ The aggregated side is the commodity.

See also: Google's support for Go, a language optimized for bus factors

> Go, a language optimized for bus factors

Can you say more about that? (I'm not much of a programmer and am not that familiar with Go.)

Bus factor: number of people with essential knowledge of the project such that things would grind to a halt if they were hit by a bus.

[I'm by no means a Go expert, but languages are my thing and I have been following Go's development.]

Go has an extreme focus on simplicity, to the extent that most of its virtues and basically all of its faults can be attributed to this.

This includes simplicity of the language itself: I learned Go incidentally, by referring to a program written in it when I was implementing an undocumented protocol. You don't need to hire people who already know the language, because it hardly matters; Go basically exists within whatever language they already know.

It also includes a homogenizing effect. Different projects written in C++ can use completely different corners of the language. Someone joining a Go project has very little "how is this project written" to deal with--in fact, the only advantages of Go I can think of that aren't rooted in simplicity are: gofmt, a tool (and more importantly, a convention) for standardizing code style; and goroutines, a standardized (and easy-to-learn) approach to the inherently difficult problem of concurrency (that otherwise has a great diversity of approaches).

So Go expands hiring pools by being so simple it's not important that candidates have experience with it, and reduces onboarding time by making projects as homogeneous as possible.

I think the key is to commoditize the complement when the complement is embedded deeper in the value chain than your product.

It works in all directions. Microsoft managed to commoditize upstack (Internet Explorer) and downstack (hardware) from its OS. The cloud vendors are attempting to commoditize the chips deeper in the stack (see Google's TPU and Amazon's new CPU thing) as well as AI tooling higher in the stack (Tensorflow). If you're a Vegas casino (and I mean, who hasn't been one at least once), you subsidize the flights to Vegas as well as the food on site. The flights get customers in the door and the food keeps them going. For the GCP, ML software gets them in the door. Cheaper processing keeps them going. Chipmakers like Nvidia and Intel are trying to do the same thing to ML software with projects like Rapids and BigDL.

Almost any layer can commoditize all around itself, as long as it has the critical mass of users. Google first got them with the search layer (who knew that was going to be such a thing in the late 90s?), which commoditized content. Netscape got it and then lost it.

Once you achieve dominance around a standard and necessary protocol at one layer of the stack, which is the whole game in tech, you can lay waste to competition in all directions, to ensure that no other layer of the stack commoditizes you.

How is Google's TPU chip a commodity? If anyone is commoditizing a ML chip, it's ironically-enough Nvidia <nvdla.org>. (And the RISC-V ecosystem will most likely get there eventually, but that's kindof a given anyway. Systolic arrays are very old tech, of Atari Transputer fame.)

Google's TPU decreases demand for chips from Nvidia, by introducing a competing chip within Google that runs ML workloads really quickly, so people pay less for algorithm training. Since TPUs are not commercially available outside of GCP, GPUs are only commoditized for Google and whoever uses TPUs on its cloud.

That's not commoditization. It's the exact opposite where Google is trying to have a differentiated TPU product they use.

It doesn't matter how TPUs are different from GPUs. It matters what the two chips have in common, which is the kind of workloads you can run on them. Any good that is a cheaper and plausible alternative to another good commoditizes the latter.

IBM originally thought that software was the complement to hardware. Nowadays, we think of hardware as the complement to software.

What does, "embedded deeper in the value chain than your product" mean in layman's terms? Does it mean something which causally precedes the product?

> Nowadays, we think of hardware as the complement to software.

We used to think that.

Apple has been very successful in reversing that again on mobile, so software is once again the complement.

"$1 for an app. Ripoff!!!"

Yes, that’s one way of putting it. In general, by ‘deeper’, I mean it’s more abstracted away from the final customer, that it is ‘further away’ from them.

For example, if you sell toasters, you better hope that electricity has been commoditized in the value chain that you operate in.

Another corollary is that you should commoditize ‘deeper’ complements when you need to defend existing money-makers when they’re at risk of bottom-up disruption. Mobile could have seriously hurt Google’s ad business, so it was a smart defensive play to commoditize mobile os on their terms via Android. To see what happens otherwise, note that they pay Apple $12B annually to have google as safari’s search default.

I mean it’s more abstracted away from the final customer, that it is ‘further away’ from them.

What I'm getting from this is: 1) in the causal chain of competitor's profits 2) better camouflaged from the POV of your customers.

Reply on news.ycombinator.com