Computed·Blg

Via Wired

-----

The ancient Library of Alexandria may have been the largest collection of human knowledge in its time, and scholars still mourn its destruction. The risk of so devastating a loss diminished somewhat with the advent of the printing press and further still with the rise of the Internet. Yet centralized repositories of specialized information remain, as does the threat of a catastrophic loss.

Take GitHub, for example.

GitHub has in recent years become the world’s biggest collection of open source software. That’s made it an invaluable education and business resource. Beyond providing installers for countless applications, GitHub hosts the source code for millions of projects, meaning anyone can read the code used to create those applications. And because GitHub also archives past versions of source code, it’s possible to follow the development of a particular piece of software and see how it all came together. That’s made it an irreplaceable teaching tool.

The odds of Github meeting a fate similar to that of the Library of Alexandria are slim. Indeed, rumor has it that Github soon will see a new round of funding that will place the company’s value at $2 billion. That should ensure, financially at least, that GitHub will stay standing.

But GitHub’s pending emergence as Silicon Valley’s latest unicorn holds a certain irony. The ideals of open source software center on freedom, sharing, and collective benefit—the polar opposite of venture capitalists seeking a multibillion-dollar exit. Whatever its stated principles, GitHub is under immense pressure to be more than just a sustainable business. When profit motives and community ideals clash, especially in the software world, the end result isn’t always pretty.

Sourceforge: A Cautionary Tale

Sourceforge is another popular hub for open source software that predates GitHub by nearly a decade. It was once the place to find open source code before GitHub grew so popular.

There are many reasons for GitHub’s ascendance, but Sourceforge hasn’t helped its own cause. In the years since career services outfit DHI Holdings acquired it in 2012, users have lamented the spread of third-party ads that masquerade as download buttons, tricking users into downloading malicious software. Sourceforge has tools that enable users to report misleading ads, but the problem has persisted. That’s part of why the team behind GIMP, a popular open source alternative to Adobe Photoshop, quit hosting its software on Sourceforge in 2013.

Instead of trying to make nice, Sourceforge stirred up more hostility earlier this month when it declared the GIMP project “abandoned” and began hosting “mirrors” of its installer files without permission. Compounding the problem, Sourceforge bundled installers with third party software some have called adware or malware. That prompted other projects, including the popular media player VLC, the code editor Notepad++, and WINE, a tool for running Windows apps on Linux and OS X, to abandon ship.

It’s hard to say how many projects have truly fled Sourceforge because of the site’s tendency to “mirror” certain projects. If you don’t count “forks” in GitHub—copies of projects developers use to make their own tweaks to the code before submitting them to the main project—Sourceforge may still host nearly as many projects as GitHub, says Bill Weinberg of Black Duck Software, which tracks and analyzes open source software.

But the damage to Sourceforge’s reputation may already have been done. Gaurav Kuchhal, managing director of the division of DHI Holdings that handles Sourceforge, says the company stopped its mirroring program and will only bundle installers with projects whose originators explicitly opt in for such add-ons. But misleading “download” ads likely will continue to be a game of whack-a-mole as long as Sourceforge keeps running third-party ads. In its hunt for revenue, Sourceforge is looking less like an important collection of human knowledge and more like a plundered museum full of dangerous traps.

No Ads (For Now)

GitHub has a natural defense against ending up like this: it’s never been an ad-supported business. If you post your code publicly on GitHub, the service is free. This incentivizes code-sharing and collaboration. You pay only to keep your code private. GitHub also makes money offering tech companies private versions of GitHub, which has worked out well: Facebook, Google and Microsoft all do this.

Still, it’s hard to tell how much money the company makes from this model. (It’s certainly not saying.) Yes, it has some of the world’s largest software companies as customers. But it also hosts millions of open source projects free of charge, without ads to offset the costs storage, bandwidth, and the services layered on top of all those repos. Investors will want a return eventually, through an acquisition or IPO. Once that happens, there’s no guarantee new owners or shareholders will be as keen on offering an ad-free loss leader for the company’s enterprise services.

Other freemium services that have raised large rounds of funding, like Box and Dropbox, face similar pressures. (Box even more so since going public earlier this year.) But GitHub is more than a convenient place to store files on the web. It’s a cornerstone of software development—a key repository of open-source code and a crucial body of knowledge. Amassing so much knowledge in one place raises the specter of a catastrophic crash and burn or disastrous decay at the hands of greedy owners loading the site with malware.

Yet GitHub has a defense mechanism the librarians of ancient Alexandria did not. Their library also was a hub. But it didn’t have Git.

Git Goodness

The “Git” part of GitHub is an open source technology that helps programmers manage changes in their code. Basically, a team will place a master copy of the code in a central location, and programmers make copies on their own computers. These programmers then periodically merge their changes with the master copy, the “repository” that remains the canonical version of the project.

Git’s “versioning” makes managing projects much easier when multiple people must make changes to the original code. But it also has an interesting side effect: everyone who works on a GitHub project ends up with a copy own their computers. It’s as if everyone who borrowed a book from the library could keep a copy forever, even after returning it. If GitHub vanished entirely, it could be rebuilt using individual users’ own copies of all the projects. It would take ages to accomplish, but it could be done.

Still, such work would be painful. In addition to the source code itself, GitHub is also home to countless comments, bug reports and feature requests, not to mention the rich history of changes. But the decentralized nature of Git does make it far easier to migrate projects to other hosts, such as GitLab, an open source alternative to GitHub that you can run on your own server.

In short, if GitHub as we know it went away, or under future financial pressures became an inferior version of itself, the world’s code will survive. Libraries didn’t end with Alexandria. The question is ultimately whether GitHub will find ways to stay true to its ideals while generating returns—or wind up the stuff of legend.

Via Tech Crunch

-----

Google, Microsoft, Mozilla and the engineers on the WebKit project today announced that they have teamed up to launch WebAssembly, a new binary format for compiling applications for the web.

The web thrives on standards and, for better or worse, JavaScript is its programming language. Over the years, however, we’ve seen more and more efforts that allow developers to work around some of the limitations of JavaScript by building compilers that transpile code in other languages to JavaScript. Some of these projects focus on adding new features to the language (like Microsoft’s TypeScript) or speeding up JavaScript (like Mozilla’s asm.js project). Now, many of these projects are starting to come together in the form of WebAssmbly.

The new format is meant to allow programmers to compile their code for the browser (currently the focus is on C/C++, with other languages to follow), where it is then executed inside the JavaScript engine. Instead of having to parse the full code, though, which can often take quite a while (especially on mobile), WebAssembly can be decoded significantly faster

The idea is that WebAssembly will provide developers with a single compilation target for the web that will, eventually, become a web standard that’s implemented in all browsers.

By default, JavaScript files are simple text files that are downloaded from the server and then parsed and compiled by the JavaScript engine in the browser. The WebAssembly team decided to go with a binary format because that code can be compressed even more than the standard JavaScript text files and because it’s much faster for the engine to decode the binary format (up to 23x faster in the current prototype) than parsing asm.js code, for example.

Mozilla’s asm.js has long aimed to bring near-native speeds to the web. Google’s Native Client project for running native code in the browser had similar aims, but got relatively little traction. It looks like WebAssemly may be able to bring the best of these projects to the browser now.

As a first step, the WebAssembly team aims to offer about the same functionality as asm.js (and developers will be able to use the same Emscripten tool for WebAssembly as they use for compiling asm.js code today).

In this early stage, the team also plans to launch a so-called polyfill library that will translate WebAssembly code into JavaScript so that it can run in any browser — even those without native WebAssembly support (that’s obviously a bit absurd, but that last step won’t be needed once browsers can run this code natively). Over time then, the teams will build more tools (compilers, debuggers, etc.) and add support for more languages (Rust, Go and C#, for example).

As JavaScript inventor (and short-term Mozilla CEO) Brendan Eich points out today, once the main browsers support the new format natively, JavaScript and WebAssembly will be able to diverge again.

The team notes that the idea here is not to replace JavaScript, by the way, but to allow many more languages to be compiled for the Web. Indeed, chances are that both JavaScript and WebAssembly will be used side-by-side and some parts of the application may use WebAssembly modules (animation, visualization, compression, etc.), while the user interface will still be mostly written in JavaScript, for example.

It’s not often that we see all the major browser vendors work together on a project like this, so this is definitely something worth watching in the months and years ahead.

Via Tech Crunch

-----

While companies like Facebook have been relatively open about their data center networking infrastructure, Google has generally kept pretty quiet about how it connects the thousands of servers inside its data centers to each other (with a few exceptions). Today, however, the company revealed a bit more about the technology that lets its servers talk to each other.

It’s no secret that Google often builds its own custom hardware for its data centers, but what’s probably less known is that Google uses custom networking protocols that have been tweaked for use in its data centers instead of relying on standard Internet protocols to power its networks.

Google says its current ‘Jupiter’ networking setup — which represents the fifth generation of the company’s efforts in this area — offers 100x the capacity of its first in-house data center network. The current generation delivers 1 Petabit per second of bisection bandwidth (that is, the bandwidth between two parts of the network). That’s enough to allow 100,000 servers to talk to each other at 10GB/s each.

Google’s technical lead for networking, Amin Vahdat, notes that the overall network control stack “has more in common with Google’s distributed computing architectures than traditional router-centric Internet protocols.”

Here is how he describes the three key principles behind the design of Google’s data center networks:

• We arrange our network around a Clos topology, a network configuration where a collection of smaller (cheaper) switches are arranged to provide the properties of a much larger logical switch.
• We use a centralized software control stack to manage thousands of switches within the data center, making them effectively act as one large fabric.
• We build our own software and hardware using silicon from vendors, relying less on standard Internet protocols and more on custom protocols tailored to the data center.

Sadly, there isn’t all that much detail here — especially compared to some of the information Facebook has shared in the past. Hopefully Google will release a bit more in the months to come. It would be especially interesting to see how its own networking protocols work and hopefully the company will publish a paper or two about this at some point.

Via io9

-----

Via TechCrunch

-----

According to the Labor Department, the U.S. economy is in its strongest stretch in corporate hiring since 1997. Given the rapidly escalating competition for talent, it is important for employers, job seekers, and policy leaders to understand the dynamics behind some of the fastest growing professional roles in the job market.

For adults with a bachelor’s degree or above, the unemployment rate stood at just 2.7 percent in May 2015. The national narrative about “skills gaps” often focuses on middle-skill jobs that rely on shorter-term or vocational training – but the more interesting pressure point is arguably at the professional level, which has accounted for much of the wage and hiring growth in the U.S. economy in recent years. Here, the reach and impact of technology into a range of professional occupations and industry sectors is impressive.

Software is eating the world

In 2011, Netscape and Andreessen Horowitz co-founder Marc Andreessen coined the phrase “software is eating the world” in an article outlining his hypothesis that economic value was increasingly being captured by software-focused businesses disrupting a wide range of industry sectors. Nearly four years later, it is fascinating that around 1 in every 20 open job postings in the U.S. job market relates to software development/engineering.

The shortage of software developers is well-documented and increasingly discussed. It has spawned an important national dialogue about economic opportunity and encouraged more young people, women, and underrepresented groups to pursue computing careers – as employers seek individuals skilled in programming languages such as Python, JavaScript, and SQL.

Although most of these positions exist at the experienced level, it is no surprise that computer science and engineering are among the top three most-demanded college majors in this spring’s undergraduate employer recruiting season, according to the National Association of Colleges and Employers.

Discussion about the robust demand and competition for software developers in the job market is very often focused around high-growth technology firms such Uber, Facebook, and the like. But from the “software is eating the world” perspective, it is notable that organizations of all types are competing for this same talent – from financial firms and hospitals to government agencies. The demand for software skills is remarkably broad.

For example, the top employers with the greatest number of developer job openings over the last year include JP Morgan Chase, UnitedHealth, Northrup Gruman, and General Motors, according to job market database firm Burning Glass Technologies.

Data science is just the tip of the iceberg

Another surge of skills need related to technology is analytics and the ability to work with, process, and interpret insights from big data. Far more than just a fad or buzzword, references to analytical and data-oriented skills appeared in 4 million postings over the last year – and data analysis is one of the most demanded skills by U.S. employers, according to Burning Glass data.

The Harvard Business Review famously labeled data scientist roles “the sexiest job of the 21st century” – but while this is a compelling new profession by any measure, data scientists sit at the top of the analytics food chain and likely only account for tens of thousands of positions in a job market of 140 million.

What often goes unrecognized is that similar to and even more so than software development, the demand for analytical skills cuts across all levels and functions in an organization, from financial analysts and web developers to risk managers. Further, a wide range of industries is hungry for analytics skills – ranging from the nursing field and public health to criminal justice and even the arts and cultural sector.

As suggested by analytics experts such as Tom Davenport, organizations that are leveraging analytics in their strategy have not only world-class data scientists – but they also support “analytical amateurs” and embed analytics throughout all levels of their organization and culture. For this reason, the need for analytics skills is exploding within a variety of employers, and analytics and data-related themes top many corporate strategy agendas.

Analytics: Digital marketing demands experienced talent

Change is also afoot as digital and mobile channels are disrupting the marketing landscape. According to the CMO Council, spending on mobile marketing is doubling each year, and two-thirds of the growth in consumer advertising is in digital. In an economic expansion cycle, awareness-building and customer acquisition is where many companies are investing. For these reasons, marketing managers are perhaps surprisingly hard to find.

For example, at high-growth tech companies such as Amazon and Facebook, the highest volume job opening after software developer/engineer is marketing manager. These individuals are navigating new channels, as well as approaches to customer acquisition, and they are increasingly utilizing analytics. The marketing manager is an especially critical station in the marketing and sales career ladder and corporate talent bench – with junior creative types aspiring to it and senior product and marketing leadership coming from it.

The challenge is that marketing management requires experience: Those with a record of results in the still nascent field of digital marketing will be especially in demand.

Social media: not just a marketing and communications skill

Traditionally thought of in a marketing context, social media skills represent a final “softer” area that is highly in demand and spans a range of functional silos and levels in the job market — as social media becomes tightly woven into the fabric of how we live, work, consume and play.

While many organizations are, of course, hiring for social media-focused marketing roles, a quick search of job listings at an aggregator site such as Indeed.com reveals 50,000 job openings referencing social media. These range from privacy officers in legal departments that need to account for social media in policy and practice, to technologists who need to integrate social media APIs with products, and project managers and chiefs of staff to CEOs who will manage and communicate with internal and external audiences through social media.

Just as skills in Microsoft Office have become a universal foundation for most professional roles, it will be important to monitor how the use of social media platforms, including optimization and analytics, permeates the job market.

The aforementioned in-demand skills areas represent more of a structural shift than an issue du jour or passing trend. It is precisely the rapid, near daily change in software- and technology-related skills needs that necessitates new approaches to human capital development. While traditional long-term programs such as college degrees remain meaningful, new software platforms, languages, apps and tools rise annually. Who in the mainstream a few years ago had heard of Hadoop or Ruby?

Each month, new partnerships and business models are being formed between major employers, educational institutions and startups – all beginning to tackle novel approaches to skills development in these areas. Certificate programs, boot camps, new forms of executive education, and credentialing are all targeting the problem of producing more individuals with acumen in these areas.

As technology continues to extend its reach and reshape the workforce, it will be important to monitor these issues and explore new solutions to talent development.