Welcome to This Week in Modern Software, or TWiMS, our weekly roundup of the most important stories and events surrounding software analytics, cloud computing, application monitoring, development methodologies, programming languages, and other issues that influence modern software.
This week, our top story concerns the spread of artificial intelligence into real-world applications.
Editor’s Note: TWiMS is taking a break next week as we focus on the really big news: FutureStack15 in San Francisco. We’ll be back on Friday, Nov. 20.
TWiMS Top Story:
Inside Facebook’s MemNets and the Drive to Bring Advanced AI to Everyday Tasks—Fast Company
What it’s about: Fast Company brings us up to speed on Facebook’s intensifying efforts behind artificial intelligence and its vast potential use cases, especially in relation to the company’s much-publicized goal of connecting billions of people in developing countries. The impetus: Facebook CTO Mike Schroepfer wrote a blog post this week detailing new milestones for the company’s AI research team, building upon an earlier demo of its natural-language processing system Memory Networks, or MemNets. According to Schroepfer, Facebook’s AI team has made breakthroughs in pairing natural language processing with image recognition (a person could ask for help identifying objects in a photo in their newsfeed, for example), predictive learning (it can beat most humans by scoring around 90% in a series of visual tests), and planning (Facebook has created an AI bot to play the game Go, which, unlike chess, checkers, and other games, has yet to be dominated by machine players.)
Why you should care: Some people might hear “artificial intelligence” and envision a dystopian future ruled by tyrannical, sentient machines. But in the near term, AI or AI-like apps will more likely help with the mundane tasks of modern life: Responding to email, making sense of information, getting from point A to point B. Fast Company’s Daniel Terdiman notes there’s no actual product in the offing from Facebook, but MemNets may very well be a precursor of a personal assistant app (à la Apple Siri, Microsoft Cortana, or Facebook M) that can understand and respond to complex natural language queries.
- Soon, Gmail’s AI Could Reply to Your Email for You—Wired
- New Milestones in Artificial Intelligence Research—Facebook Newsroom
- Facebook Aims Its AI at the Game No Computer Can Crack—Wired
- Apple’s Secrecy Hurts Its AI Development—Bloomberg
What it’s about: Earlier this year, retail giant (and FutureStack15 presenter) Target announced it would invest $1 billion in technology as part of a broader, company-wide transformation. The move is already underway, with Target rebuilding its engineering teams and culture from ground up and moving its DevOps adoption out of the shadows and into the highest levels of the company. The story of Target’s tech makeover under CIO Mike McNamara, first reported by CIO Journal’s Rachael King and later featured over at DZone, is essentially a real-time, developing case study in digital transformation and the expansive impacts of modern software. (And if you come to FutureStack15 Nov. 11-13 in San Francisco, you’re sure to hear more about it.)
Why you should care: Conventional wisdom holds that large, traditional enterprises are slow-moving and hesitant to change. There’s some truth in that, but when it comes to DevOps, Target’s effectively saying: We’re gonna do it anyway. DZone MVB Yaniv Yehuda recaps some of the highlights of King’s story, including Target’s “Dojo,” an internal incubator for training engineering teams in how to make the most of a DevOps model, as well as methodologies such as agile development, while completing 30-day challenges in two-day sprints. This isn’t just a corporate edict handed down to the rank-and-file: “A year ago, people wouldn’t even use the word DevOps and now it is part of daily conversations in the hallways,” King quotes Target senior group manager Ross Clanton. Example: Senior executives are participating in DevOps workshops that require them to push code into production. Developers who thought the bosses have no idea what’s required in a production deployment should enjoy that one.
- Target Rebuilds Its Engineering Culture, Moves to DevOps—CIO Journal
- 10 Reasons to Attend FutureStack15—New Relic Blog
The Founders of $3 Billion Atlassian Won’t Be Happy Until 100 Million People Use It Every Month—Business Insider
What it’s about: Atlassian recently hit a goal co-founders Mike Cannon-Brookes and Scott Farquhar set when they first set out in 2002: 50,000 paying customers across its various enterprise apps for software teams, which include JIRA, BitBucket, Confluence, and HipChat. But as they tell Business Insider, they’re eyeing a much loftier goal: 100 million monthly active users. “It’ll probably take another 13 years to get there,” Cannon-Brookes tell BI. Indeed, that’s an ambitious number for an enterprise software firm, especially one with a core focus on dev teams. But it clearly speaks to the firm’s patience and apparent lack of interest in being an overnight sensation.
Why you should care: Atlassian has plenty to teach founders and developers building enterprise apps. Business Insider points out a couple of near-astonishing facts: The company has never raised venture capital, and it has never employed a salesperson. And, unlike some other so-called unicorns (privately held firms valued at a billion dollars or more), Atlassian is already profitable as it approaches a potential IPO. In fact, when you dig into Atlassian’s story, the company begins to sound like Exhibit A for Sam Altman’s Startup Playbook (see below): It tackles real business problems, and it strives to operate with a relentless focus on product quality and continuous, incremental improvement. As Matt Weinberger writes: “Each individual improvement may not seem like much, but it makes it that little bit more appealing to a customer who’d otherwise not want to take the leap.”
Startup Playbook—Sam Altman
What it’s about: Y Combinator president Sam Altman has advised more than a handful of startups, and there are plenty of common denominators in the startup success stories he and the YC partners see in their companies. Sure, every startup is (or at least should be) different, but why not collect fundamental advice given to YC companies and share it with everyone? That’s just what Altman has done with his Startup Playbook, an insightful, illustrated guide to the things he thinks that startups must do to succeed. It’s long, but definitely worth a read for founders, would-be founders, and anyone interested in the startup ecosystem. Altman covers everything from building a team to fundraising to the role of the CEO, but he focuses on the one thing that he sees as separating successful startups from those doomed to fail. Frankly, one point holds true of software at any stage: Build a great product. “If you do not build a product users love, you will eventually fail,” Altman writes. “Yet founders always look for some other trick. Startups are the point in your life when tricks stop working.”
Why you should care: It sounds so simple, doesn’t it? So how do you do it? Altman advises a “product improvement engine” approach: Get as close to users as possible (“Sit in their office if you can”) and watch them use your product. (Altman shares that Pinterest’s Ben Silbermann used to ask strangers in Palo Alto coffee shops to try an early version of the site.) Figure out what’s missing or not working well enough, and improve it. “If you improve your product 5% every week, it will really compound,” Altman advises. “The faster the repeat rate of this cycle, the better the company usually turns out.” Eventually, happy users become your best salespeople (see Atlassian’s story above). Altman also highlights an oft-overlooked distinction between building consumer and enterprise apps. If you’re working on a consumer product, launch as quickly as possible, he writes, and then recruit people to use it. But if you’re building an enterprise product, he advises finding a paying customer before you start writing code: “Talk to potential customers and then figure out the smallest subset of the technology you can build first.”
What it’s about: Silicon Valley’s diversity problem is significant. As a group, 101 industry leaders in The Atlantic’s second-annual Silicon Valley Industry Insiders poll pegged the problem at a little higher than seven. They rated sexism in tech roughly the same. And 17% of respondents indeed find the lack of racial and ethnic diversity in Silicon Valley unfathomably bad, rating it 10. Microsoft researcher Kate Crawford asked: “Can your scale go up to 11?”
Why you should care: It’s easy to make excuses or simply look the other way, especially when the industry is thriving. “We must be doing something right,” the thinking goes. Waldo Jaquith, director, U.S. Open Data, would like to dispel that notion: “The entire industry falls back on [B.S.] excuses like ‘It’s a pipeline problem’ and ‘We’re a meritocracy,’” Jaquith says in a quote in the poll’s results. It’s a “pernicious,” self-serving logic, he says. The Atlantic’s Robinson Meyer notes education and labor stats that undermine the pipeline-problem excuse, in particular: Black students earned 4.1% of the bachelor’s degrees in computer science granted by elite U.S. and Canadian universities during the past several years, according to a recent poll. “Yet blacks only make up about 2% of recent hires at tech firms,” Meyer writes.
- Twitter Engineering Manager Leaves Company Because of Diversity Issues—TechCrunch
- Silicon Valley Struggles to Hack Its Diversity Problem—The Washington Post
Microsoft Used to Threaten to Send Weak Programmers to Work on a Bizarre App About Dogs—Business Insider
What it’s about: Microsoft posted an hour-long video tour of its archives this week. In it, veteran Windows engineer Raymond Chen shares one of management’s fear-as-motivator tactics from back when the development team was building Windows 95: If the bosses were unhappy with your performance, they threatened to reassign you to work on a slightly lower-profile software project: Microsoft Dogs. It was not an empty threat, but to call it an “app” might not do the time period justice: Microsoft Dogs was a very real CD-ROM released in 1995 (here’s its review in Entertainment Weekly) that, as Business Insider notes in its recap of the video, “the company made to inform dog owners of the proper care and feeding for 500 different breeds of dogs.”
Why you should care: In the same video, Microsoft employees discuss the notorious failure of Windows Vista and wonder—while acknowledging flaws in Vista’s development—whether the operating system would have been better received if it was released a couple of years later. Well, imagine how the poor Microsoft Dogs team must feel. Tell the right person in Silicon Valley you’re working on an app for dog owners today, they’ll probably sign a term sheet for a few million dollars in seed funding.
- Defrag Tools #143: Raymond Chen – Microsoft Archives Part 1—Microsoft Channel 9
Want to suggest something that we should cover in the next edition of TWiMS? Email us at firstname.lastname@example.org.
Tune In to the Future
Can’t get enough modern software news and commentary? Be sure to check out our new Modern Software Podcast. New Relic Editor-in-Chief Fredric Paul and guests discuss the most important things happening in the world of software analytics, cloud computing, application monitoring, development methodologies, programming languages, and more. Listen to episode 2 or subscribe on iTunes.