Joel Spolsky – Joel on Software

I’m Joel Spolsky, a software developer in New York City. More about me.
Ready to level up? Stack Overflow Jobs is the job site that puts the needs of developers first. Whether you want to take control of your search or let employers discover you, we’re on a mission to help every developer find a job they love.
Looking to hire smart programmers who get things done? Stack Overflow Talent is a fully-customized sourcing solution that helps you understand, reach, and attract developers on the platform they trust most. Find the right candidates for your jobs. Learn more.
For my day job, I'm the co-founder and CEO of Stack Overflow, the largest online community for programmers to learn, share their knowledge, and level up. Each month, more than 40 million professional and aspiring programmers visit Stack Overflow to ask and answer questions and find better jobs. Stack Overflow is also the flagship site of the Stack Exchange network, 160+ question and answer sites dedicated to all kinds of topics from cooking to gaming. According to Quantcast, Stack Overflow is the 30th largest web property in the United States and in the top 100 in the world.
I also founded Fog Creek Software, one of the most influential small tech companies in the world. As an independent, privately-owned company, we’ve been making customers happy since the turn of the century. We share what we've learned about how to make great software, both by writing about our ideas and by creating products, like FogBugz, Trello and Gomix, that help others make great technology. As a result, Fog Creek's impact on the world of developers rivals companies a thousand times our size.
Since the 1990s, the web has been a publishing place for human-readable documents.
Documents published on the web are in HTML. HTML has a little bit of structure, for example, “here is a paragraph” or “emphasize this word.”
Then you stir in some CSS, which adds some pretty decorations to the structure, saying things like: make those paragraphs have tiny gray sans-serif text! And then people think you are hip. Unless they are older, and they can’t read your tiny gray words, so they give up on you.
That’s “structure,” as far as it goes, on the web.
Imagine, for example, that you mention a book on the web.
Goodnight Moonby Margaret Wise BrownIllustrated by Clement HurdHarper & Brothers, 1947ISBN 0-06-443017-0
There’s not much structure there. A naive computer program reading this web page might not realize I was even mentioning a book. All I did was make the title bold.
So, also since the 1990s, people have realized that we can make the web a much more useful place to publish information if we applied a bit more structure. As early as 1999, Tim Berners-Lee was writing about the Semantic Web:
“I have a dream for the Web [in which computers] become capable of analyzing all the data on the Web – the content, links, and transactions between people and computers. A ‘Semantic Web’, which makes this possible, has yet to emerge, but when it does, the day-to-day mechanisms of trade, bureaucracy and our daily lives will be handled by machines talking to machines. The ‘intelligent agents’ people have touted for ages will finally materialize.”
Using the Semantic Web you might publish a book title with a lot more detail that makes it computer-readable. To do this, you would probably start by going to schema.org and looking up their idea of a book. Then you could use one of a number of formats, like RDF or JSON-LD, to add additional markup to your HTML saying “hey! here’s a book!”
Ok, well, doing that is kinda hard to figure out, and, to be honest, it’s homework. Once your beautiful blog post is published and human-readable, it’s hard to gather the mental energy to figure out how to add the additional fancy markups that will make your web page computer-readable, and, unless there is already a computer reading your web pages, at this point, you usually give up. So, yeah. That was 1999, and not much progress has been made and there is very little of this semantic markup in the wild.
We would like to fix this, because human progress depends on getting more and more information in formats that are readily accessible, both by regular humans, their dumb A.I. li’l sibs, and your more traditional computer programs.
Here is something I believe: people will only add semantic markup to their web pages if doing so is easier than not.
In other words, the cost of adding semantic markup has to be zero or negative, or this whole project is not going anywhere.
Now imagine this world for a second:
In this world I did less work to insert a book (because I was assisted by a UI that looked up the details for me).
You can imagine the same scenario applying to literally any other kind of structured data.
My “address block” might have any visual appearance. Visitors to my web page might see the address, or a little map, or a little map in Japanese, etc. etc. The semantic content is there behind the scenes. So, for example, my web browser might know “gosh this is an address! Maybe you want to do address-y things with it, like go there,” and then my browser might offer me options to summon a self-driving car and even call an ambulance when the self-driving car self-drives into a snowbank.
My two simplistic examples of “book” and “address” are interesting right now because (a) you can probably think of 1,000,000 more data types like this, and (b) none of these things work right now, because even though almost every web editing environment has a concept of “blocks,” none of them are extensible. WordPress has (oh gosh) hundreds of block types, but they don’t have thousands or millions, they don’t have “book” or “address” or “Burning Man Theme Camp” yet, and there’s no ecosystem by which developers and users can contribute new block types.
So I guess I gotta wait around for someone at WordPress to develop all the blocks I want to use. And then someone at Notion, and then someone at Trello, and then someone at