Let’s stop writing like it’s 1995

Jensen Harris
May 4, 2016 · 10 min read

It’s not every day when a technology is created that fundamentally changes writing. Or even every year.

In fact, looking back over the history of writing on computers, there are only a few real inflection points at which the confluence of new technology, personal computing power, and user interface design have made the process of writing drastically better.

Today, Textio has released a milestone that we believe marks one of these rare inflection points.

It has always been the goal of Textio to improve the effectiveness of written content. To do that, we started off in the talent recruiting space and compiled millions of data points around job ads. Then we applied machine learning and natural language processing techniques to see if we could learn why they performed well or not. And it worked.

Now we can take it to the next level. Textio can transform your writing — not the act of creating or editing a text-based digital document, but actually how you write.

But before I explain how, let’s consider for a moment how much technology has helped to improve the process of writing over the last 35 years.

Codes and keys: Early writing software

In the early 1980s, text-based word processors really came into their own and launched a writing revolution on personal computers. WordStar for DOS dominated the market with its arcane formatting codes and hard-to-use features. Despite nearly impenetrable user interfaces, these text-mode word processors were an immense success. Why?

WordStar for DOS

Because for the first time, from the comfort of your home or office, you could format and print text for yourself, by yourself, on a computer small enough to fit on your desk. It was a revelation. And it foretold the power of the coming personal computer revolution.

Still, there was much that these early word processors simply couldn’t do. In particular, they did virtually nothing to actually improve on the substance of your writing. There was no spell check. You couldn’t see what your work would look like once it was printed. It was just type, format, print, and pray.

The graphical user interface: a game-changer

With the introductions of the Apple Macintosh in 1984 and Microsoft Windows in 1985, the entire word processing category was blown asunder by a new disruptive technology: the graphical user interface (GUI). For the first time, you could actually see what your document looked like while you wrote it.

Bold was bold! Italics were italic! Tables!! Colors!!

Microsoft placed the biggest and earliest bet on the GUI revolution, and Microsoft Word became the most popular writing platform for the next decade.

Word came with a spell checker early on (as did the inevitable clones made by other companies) although it was universally implemented as an unwieldy dialog box that you had to navigate in order to choose a correction for every individual misspelling. This wasn’t because companies wanted to make it difficult to correct errors, it was just because computers weren’t powerful enough to have “always-on” spell checking.

The universally horrible Word spelling dialog box

Then In 1995, Word added a new feature many of us take for granted today: red squiggles under misspelled words.

Word 95 added on-the-fly spell checking

Without popping up an inconvenient dialog box, you could see what was misspelled in real time right as you typed. It was amazing! Computers were finally fast enough to actually help you write better. The power of the personal computer was at your fingertips.

Writing in the clouds

A decade later, in 2006, Google acquired one of the first web-based word processors, Writely, and turned it into Google Docs. The internet age of writing was born, enabled by two transformative technologies: the internet itself, and sufficient network connectivity to ensure that people could assume they were always online. We think of those collectively today as the cloud.

Writely became Google Docs in 2006

The internet age deemphasized many of the rich formatting controls that seemed so important in the GUI era in favor of features built to take advantage of the connective power of the web. Features like writing within any web browser with no software install, sharing a universal link to a document, multiuser editing, and online commenting and collaboration upended the world’s expectations of productivity software and how it worked and looked.

And yet, one thing didn’t change: None of these new features actually made your writing fundamentally better or more effective. While Google and Microsoft continued to pack more widgets and capabilities into your text editor, when it came to your actual writing, you were still pretty much on your own.

Predictive engines: the next writing revolution

Fast forward another decade to 2016. Textio, a computing platform that we launched last summer, is built on top of a remarkable, revolutionary enabling technology: machine intelligence.

For the first time in 35 years, a set of disruptive technologies and user interface breakthroughs make it possible to actually improve the substance of what you write, as you write. You are no longer alone in the white box.

Processing an enormous data set with complex algorithms that overlay artificial intelligence, machine learning, natural language processing, data science, and a rich learning loop, Textio is able to predict with high statistical accuracy how successful your written content will be when you put it out in the real world. And it does this as you write it, in real time. At Textio, we’ve come to call this a predictive engine.

Textio, the AI-based writing platform

A predictive engine considers everything about what you are writing in context — the type of content, who you are trying to reach, what industry you work in, even where in the world you are located — to give you personalized, accurate predictions of what language will cause the people you are writing for to respond more positively, or more negatively.

But unlike most supercomputers, you can tap into the power of Textio’s predictive engine just by typing, the same way you always do. It’s a personal supercomputer. Our first product, Textio Talent, is made for finding and hiring the best talent available. It helps you write the best job listings and the best recruiting emails to attract great people to your company.

Textio’s predictive engine uses outcomes data to find the language that works

Our current models are built on the outcomes data of over 15 million job listings. It is hard to overstate what a challenge it is to create a user experience where, from the time you stop typing, we have only 300ms to send your document to our servers, run all of the predictive algorithms, and then bring the predictions back to the editor. Yet, that’s what’s required to make the user experience feel optimally responsive — to make it actually feel like you have a supercomputer at your fingertips.

Language changes, and we change with it

A year ago, if you wrote “big data” in a job listing, software engineers would have been beating down your door asking for a job. About six months ago, it started to lose its luster. Today, the term “big data” is statistically an actual negative that will push engineers away from applying to your jobs. Language, like fashion, is always moving forward, and without the kind of always-on predictive engine the Textio platform is built on, you will be stuck wearing the acid washed jeans.

This is where rules-based writing checkers really start to fall apart. Who decides when a certain phrase is no longer trending? Who decides when to change the rule? A predictive engine is continually evaluating new job listings against new outcomes data, and new algorithms find the newest language that is attracting the very best candidates today, not two years ago, not even two months ago.

The best candidates in any industry get taken off the market extremely quickly (if they are ever on the market). They know they are in demand, and they respond to only the most compelling invitations. Statistically, chances are you are not even seeing these people come through your door for interviews.

Imagine having access to real-time intelligence that tells you exactly the language these exemplary candidates have responded to. In your industry. In your location. Textio customers are finding that they fill their jobs on average 17% faster than before.

The new word checkers: just like 1995

Look elsewhere in the industry and you’ll find a number of fun tools available to assess your writing, each checking different aspects of your words. Instead of misspellings, they try to spot emotional sentiment, or verify good grammar, or grade your reading level.

Unlike Textio, these tools generally operate on much the same basis as 1995-era spell check: a relatively small and static list of rules is used to test your words and then mark them up, much as an elementary school teacher might use a big red pen all over your essay.

These tools are fun to try out, but their real-world impact on writing for business is fairly limited. Since they are general-purpose writing tools, they can only offer generalized superficial advice. The specific language you use to write a job description vs. a sales email vs. a performance review vs. a marketing campaign vs. a 1:1 exchange is vastly different. So these add-ins can offer only a limited scope of help.

It’s chess, not spell checkers

I started this post by promising to tell you about the newest technology we’ve built here at Textio. Today, we are announcing a monumental step forward: something we call opportunities.

Before today, Textio’s predictive engine worked like this: you wrote things, and Textio would tell you which of those things were working, which weren’t, who would respond favorably to your writing, and what exactly you needed to change to attract the people you wanted to attract. Ridiculously powerful and essential. But we knew we could do more.

What occurred to us was this: it isn’t enough to mark up what you’ve already written. What we need to do is imagine the things you haven’t written and tell you which one of those alternate universes contains the best version of your writing.

It turns out this is the kind of thing supercomputers are really good at. The world was amazed in 1997 when an IBM supercomputer called Deep Blue defeated reigning world chess champion Garry Kasparov. Just this year, Google’s AlphaGo supercomputer beat Go world champion Lee Se-dol three times in a row. Go was thought to be the last remaining strategy game in which humans could still beat machine intelligence.

What we’ve done with Textio’s predictive engine is conceptually similar to a chess supercomputer: instead of looking at just the chess pieces on the board and grading your moves, we look hundreds or thousands of moves ahead, finding the most effective phrases that you didn’t write… but could have. The most impactful of these are surfaced as “opportunities.”

Textio “opportunities” help find the critical phrases buried deep below your text

Opportunities are shown as a green outline in Textio. In the example above, there’s nothing at all wrong with the phrase “focused on.” It neither helps nor hurts the effectiveness your content. Same with “helping people,” “helping people find,” “helping people find a lawyer,” “find a lawyer,” or the other thousands of phrase combinations you could mess around with in this sentence.

It turns out, though, that buried deep below the phrase “focused on” is a secret: a different phrase that, if used in its place, in this kind of recruiting mail, in your city, would make your job listing statistically more successful at attracting candidates to respond. It’s as if you have a supercomputer ally feverishly trying all the variations of your sentences, nudging you towards the best version of your writing.

It’s here now

Textio is pioneering the use of machine intelligence to improve writing. Most of what businesses produce all day long is writing, so this turns out to be among the most valuable technologies a company can invest in to stay ahead of their competitors.

Textio is a game-changer in the highly competitive recruiting market. No forward-looking company can afford to be without it. You will consistently be beaten out for the best and most diverse set of job candidates by competitors who are using Textio.

Here’s a short 3 minute video that shows all of this new technology in action:

We’ve begun by tuning our predictive engine for writing job listings and recruiting mail, because in the end, what is a successful company if not amazing people? But the platform can do so much more.

We’re just getting started. We believe it will change the way you write, forever.

Learn more about how language impacts your hiring at textio.com


(Thanks to Walt Mossberg, Steven Sinofsky, Gordon Ritter, Aileen Lee, Roy Bahat, Kent Goldman, Shivon Zilis, and many others who provided valuable feedback on early drafts.)

Textio Blog

Writing about the future of writing

Jensen Harris

Written by

Co-founder & CTO of Textio

Textio Blog

Writing about the future of writing

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade