April Fool The Register seeks full-time developers and data scientists to build software tools to generate El Reg-style headlines and other text.
This technology will draw upon the work of top UK academic Professor Warren Rice, who has assembled a database of more than one million news stories, including the entire Register archive, and used it to train neural networks to create headlines that draw in netizens to click on links.
Successful applicants will work with Prof Rice whom we have hired to head this development effort.
For two decades now, The Register has published online, translating news and technical knowledge into tens of thousands of articles read and enjoyed by millions globally.
Amid all those exclusives, features, and bootnotes, one aspect of our journalism has stood out. The headlines. Love them or hate them, they are one of our USPs. They have been quoted in court to derail the schemes of Apple, Intel, and other multibillion-dollar corporations. They have been read out on air by baffled BBC presenters.
And they have been printed out and stuck on office walls. Mainly as dartboards.
For years, these headlines, sub-headlines, captions, and recently the experimental next-generation of headline – tweets – have been crafted painstakingly by hand. Specifically, the hand of our small but highly skilled workforce of sarcasm artisans. These laborers, using closely guarded expertise passed down from generation to generation, use keyboards and mice to fuse computer engineering knowhow, puns, and functional alcoholism to form our article titles.
However, times are changing. Just as our cousins in the newspaper world had to ditch their unionized hot-metal printers as well as other follies such as accuracy and dignity, we too must look to the future rather dwell on the past.
It is difficult and expensive in this day and age to find, recruit, train, and retain human headline writers who know their ASCII from their elbow, are up to date on the trends triggering the hashtag generation, and can keep coming up with fresh Super Cali Goes Ballistic headings.
To modernize our digital output, rather than rely on maintaining a team of oxygen-consuming wordsmiths, we are building artificially intelligent software to create precision-engineered headlines from articles.
And as such, we are seeking programmers and data scientists to develop this critical real-time technology that will produce compelling and hilarious titles.
"This is an exciting time, and going forward an ideal opportunity to transform our staff as well as our readers, shareholders, advertisers, and mandatory regulatory framework partners," said Register editorial engineering manager Jessica Elliott.
"Right now, we're writing headlines for people already in the industry. But what about Tu Morrow on Instagram, or Jermajesty on Snap? Are we writing effective headlines for the cloud generation, for people who will never have to install an application in their life? Can a bot program outperform a human writer, and churn out mad dramz the next decade's IT workers can relate to? We think: yes.
"And if the headline-writing bots prove to be as big a success as we're targeting, who knows what's next? Answer: computer-generated articles, or comagrticles as we have already trademarked."
To score the job, you will need five or more years’ experience working with Perl, TensorFlow, and Python, plus familiarity with recurrent neural networks, Keras, and the United Kingdom's esoteric libel laws. And no, that's not a typo. Perl is because the site is powered by Perl.
Applicants must also display high-level punning and rhyming mastery, familiarity with the difference between alliteration and tautological tautologies, and the ability to quote at length the minor works of Monty Python or Ed Sheeran, or whatever the youth of today are following.
Editorial staff are vehemently opposed to this project, so the successful applicant will require superior interpersonal and conflict resolution skills. The positions are open worldwide. To apply, decrypt the following string into a valid @theregister.com email address, and send us your CV and full source code and binaries of similar work.