The key is I distinguish between an API which IS the business - something like a Twilio for example - and an API which lets you leverage a business's other assets in a way that might one day be cut off - like a Twitter.
You are safe to build on top of something like Twilio because the service is the API and API consumers are the customers. Consumers of Twitter's API are not customers of Twitter and should not be surprised Teotter does not treat them as such.
You are either the customer or the product. If you are neither you are in a precarious position.
In startups, we often tell each other that "the idea is useless", that is, startups are not about the technical details of the idea, they're about the human-to-human interaction of people and your business.
This is also true about data. Startups are not about the data. A dozen companies did what Dropbox did, but there was only one Dropbox. Everybody did search, but there is only one Google, and so forth. Yes, people consume the data, and yes, if you ask them they'll say that's why they're using your app, but that's confirmation bias. They already use your app and enjoy it. When you ask them, they just look at what the app does and repeat it in a way that makes them sound the smartest.
So, presto chango, adding those two things together along with Jacques' essay and we end up in a really weird spot: your startup may be about providing some kind of massaged data to end users. So you find an API to do that. But the guys providing the data, the heart of your app, have the least amount of reward in the entire equation! Hell, in many cases they're providing data willy-nilly out to just about everybody for free. To them, the data is worthless, yet you're building your business off them! Crazy world.
I have a personal news aggregator I wrote a couple of years ago, http://newspaper23.com Nothing fancy, just rips some headlines, then rips the plain text, then puts the plain text in json and makes a client-based detached reader.
To provide the plaintext, I was using ViewText. Great folks. Put in an URL and it'll give you a readable version of what's there.
But guess what? The ViewText guys didn't like my app ripping 50-60 articles a day. So I lost this huge amount of functionality when they decided to start rate-limiting. Now when you click an article, I bring up a frame and load it. I wanted plain text, this was the entire reason for the app, and now the critical thing I wanted is the exact thing that isn't working.
No big deal for my newspaper23 app, but it would be a freaking red alert if this was something I was selling. API use, like walled gardens, are high-risk.
Assuming that Jacques Mattheij is using the French pronounciation of his first name, the possessive version of Jacques is Jacques's.
This is because in French the final letter of most words is silent, and in English the possessive form of spoken nouns not ending in an audible "s" is rendered in text as <singular>'s.
Thank you. I changed it to the Americanized form of possessive for words ending in s (pronounced or not) I think this could go either way, but I screwed it up no matter which choice an author might take.
Still, I prefer my rendition. The problem here is the incompatibility between the rules of pronounciation -- French permits silent letters to end a word, English in general does not. In translating to English you must choose either to honour the French or the English rules of pluralisation and possession.
Because I don't write many English sentences with de la, du or d', I figure that sticking to the English rule is the best move.
Well in all fairness, outside of La Francophonie, "Jacques" is not a common name. And for English speakers it's just chock full of exciting traps. My favourite is putting 'c' and 'q' next to each other. In French that's a single sound; in English it says "new syllable, please!
edit:
> Jacqueses! (Jacuqi?)
Jacqueses would be closest (-ii is a Latin pluralisation, Jacques comes from Jacob which is Hebrew).
There's an impedance mismatch here: English stores plurality in the word, French stores it in the sentence.
"Startups are not about the data. A dozen companies did what Dropbox did, but there was only one Dropbox. Everybody did search, but there is only one Google, and so forth."
Maybe I'm missing something, but DropBox was just a very well-implemented GUI for backing up files - there was no "special sauce data".
As for Google, one could argue that their data (or their index) was one of their competitive advantages. They started crawling pages back in 1998. They basically have a 15 year head start over any new competitors that want to enter the field today. And their index is vastly superior to Bing's/Yahoo's.
So I would disagree that startups are not about the data. For a lot of them, without the data, they'd just be a fancy, useless CRUD GUI.
I always liked the term 'Business Objects' - but that has always been co-opted by a company (SAP).
So hard to invent catchy new terms... Trying here but they all sound like business buzzwords...
Business Proxy, Business Delegate, Business Front, Business Agent, Company Blackbox, Business Interface, Company Platform, Company Rube Goldberg, Company Socket, Partner Interface?
I know it's nitpicky, but the error makes it painful to read the article. Obviously I'm blaming nobody, and to judge from his name the author is probably fluent in more languages than me. But it's a user-friendliness thing. It would help.
I think you may have broken something while doing this?
I see this on your main page:
---
The Api Paradox
APR 11TH, 2013
– layout: post title: The API Paradox date: 2013-04-11 11:00 comments: true categories: api development startups programming
English is my third language(after Arabic and French) and that was still annoying to me. I know it's a nitpick, but I can't help being bothered by misplaced apostrophes.
Great writeup. Agree with everything that was said. I am currently building a service with much of the functionality based upon an API. The concerns the author has mirror my own. Perhaps there should be an API code of ethics; somewhat like open source licensing. Not one for regulation but it would be nice to have some assurances for both parties involved.
I don't think the age of the company matters much. Incentives can be aligned/misaligned at any age. I wrote a similar post not that long ago that argues many of the same points (albeit with a more incendiary title) http://thenextweb.com/dd/2013/03/12/apis-are-dead-long-live-...
I don't see why "It is possible to create a competing business using that very same Api" is a potential warning sign. Anyone got some thoughts about that?
I always thought "protocol" was a more appropriate term for such internet-based APIs. SMTP and HTTP are clearly protocols, and even APIs built on top of HTTP are generally referred to as protocols in the standards world. For example WebDAV (RFC4918) and AtomPub (RFC5023) are both considered protocols.
It seems to be the accepted use these days. I disagree with that use but I don't have a better term and don't want to add to the confusion. But you're absolutely right that an exposed service and an API are different in many fundamental respects. I think the confusion is rooted in the gray area called remote procedure calls, if you implement your service exposure as a remote procedure call (and many of the early ones were done in just that fashion) then the difference is almost moot.
Maybe WSI would be a better moniker? (Web Service Interface)?
why? for pointing out that, in 2013, someone had the revelation that if you fish in someone elses pond, be careful... because they might cut off your access?
You are safe to build on top of something like Twilio because the service is the API and API consumers are the customers. Consumers of Twitter's API are not customers of Twitter and should not be surprised Teotter does not treat them as such.
You are either the customer or the product. If you are neither you are in a precarious position.