Testing a tag manager tag

I'm writing this post to see if +Justin Cutroni's article really works. It should record a click event every time someone clicks on his google+ page or his article.

Should you pay for a MOOC-based certification in R?

My simple response is no. Instead, you should use the portfolio you create as part of your learning to aggressively market yourself to people who care about the kinds of things you can do for them. That's where you will get the payoff, and it's the only proof that really matters.

This said, I have occasionally paid for certifications. Why don't MOOCs teaching R, even very good ones, pass the test?

• Paying does not entitle you to premium support or any support really. At most, it seems you pay just to guarantee that it's really you taking the course although, to be fair, this verification may itself have some incidental value. For instance one study group of students taking Coursera's Data Science Specialization use whether you are a paying, identity-verified course participant, as one criterion for joining them. If that's important to you, then paying might be worth it.
• Employers don't really value certification unless the field is well understood and the certification tests can be used to differentiate ability in the market place. Learning R is of some value, but it's really being able to structure problems and derive useful answers that is the real value. MOOCs don't teach that, and MOOC certifications don't demonstrate you can do that.
• The quality of MOOCs is uneven. For instance, even the excellent MITx MOOC I'm currently auditing, The Analytics Edge, has some glaring mistakes in one of the problem sets. If you've taken enough courses on quantitative analysis, you know that mistakes are bound to happen. The problem is that you have no real recourse other than complaining on the forums. Why pay for that?

So, in short, I say don't pay because it doesn't get me anything. When might I pay? If I could pay $300 or $400 and get real support, I'd consider it for certain topics. I might (and have) paid several thousand dollars for learning experiences that then give me access to verifiable earning opportunities. I've also paid up to $1000 for subscriptions to services that provide real analytic value.

Pursuing Data Science in the Cloud (is it really just about R?)

I'm currently enrolled in two MOOCs for learning data science. They are:

• The Analytics Edge, from MITx (hosted on edX).
• The Data Science Specialization, from Johns Hopkins (hosted on coursera).

Let me reveal my biases. First, I vastly prefer The Analytics Edge, the MITx offering, but I don't dislike the Johns Hopkins Data Science Specialization. The Analytics Edge is very intellectually engaging while the Johns Hopkins specialization focuses on building your professional portfolio so that you can advance your career. In particular, the Johns Hopkins specialization injects an element of social networking through git and git hub that is just totally lacking in The Analytics Edge.

You need both, so I'm taking both.

Second, both offerings essentially use the statistical programming language R as their centerpiece. I have used this language extensively in the past. In the few years since I last took it up, the associated tools have taken a nice leap. In particular, there is now RStudio, a much friendlier editing and interactive debugging environment as well as the two courses I mentioned. You can now see a real practice-based community developing around the language, smoothing the rough edges it was born with as an academic brain child.

Now for the kicker, data science cannot just be about programming or understanding analytical paradigms for solving problems. I'm taking these two courses to determine the current zeitgeist. But, I think the real value in data science comes from understanding the problem to be addressed in its natural context and going from there. In other words, the critical element is in initially figuring out the problem. That critical step is absent from both courses. I think there is an opportunity there.

Big changes to Google+, Have they broken Google+ blog commenting

At 6:55 PM, May 15, 2013, Google+ blog commenting seems broken.

Feedly

The announced July 1st shutdown of Google Reader this past Wednesday sent me into a bit of a tizzy. Google Reader is a very convenient way to track web site updates. I've been using RSS, the underlying technology for Google Reader, since it was introduced in the late 1990s.

RSS originally stood for Really Simple Syndication. The idea behind RSS is to make a web site's content available for aggregation by any program capable of reading RSS. Obviously, sites may not want to syndicated full articles but only article extracts or headlines. You find all three.

RSS is an incredibly efficient way of tracking the content of multiple web sites, and Google had really invested a lot over the years improving its infrastructure to make its own tracking of RSS as efficient as possible. Google Reader was a nice side-effect of that effort.

But, the truth of the matter is that Reader's interface had become neglected over the past couple of years. Google had even resorted to ripping features out of Reader in favor of Google Plus, their flagship social network.

Google Plus doesn't really hack it for tracking website updates.

So, the past couple of days I've been evaluating non-Google alternatives to Reader. The one I like so far is Feedly. It uses Google's feed infrastructure, something that will have to change before Google shuts it down.

Here's a list of the feedly positives:

• Very fast loading.
• A compact, attractive display of headlines. It just hammers every other alternative with this feature alone.
• Great iOS apps that really facilitate skimming headlines and quickly accessing content. It's a joy to use on iOS.
• Well implemented sharing in the web interface, with a bit of an exception for Google Plus.

Here are some areas where feedly needs work:

• Sharing to Google Plus from iOS. Feedly defaults to the web interface. To be fair, Google has only recently opened up the API for doing this directly without having to go through the web.
• Sharing to Google Plus is also slightly flaky from Feedly's web interface. The text box for inputting your commentary has a tendency to flicker out.

Things any of these feed readers will have to do to attract me long term:

• Convince me they have a business model. For instance, how does Feedly make money? Unclear.
• Provide me a way to leave by exporting my feeds.

Why not a web book vs. a chromebook

This post is inspired by +John Hardy's rant today concerning something called asm.js as well as what I will only describe as long term recall of +Tim Bray's native vs. web article of a couple of years ago.

My key point is this: Google's Chromebook experiment has it right on how people spend their time on light weight laptops but wrong on what to do about it. In short, as assumed by the Chromebook team, people on light weight laptops do spend the vast majority of their time interacting with and publishing "web" content.

However, is the solution to just push everything into the browser? I'll note that much like with my mobile devices, I opt for native web clients on my macbook air when they're available. They're faster, and I have more control over each service.

I think the Chromebook would be more interesting if it more clearly recognized this fact. Perhaps Go could become the default native client language.

The pixelated cloud

I know +Sundar Pichai said that Google's new chromebook is named Pixel because we'll never have to consider pixels again once we start using it to access the cloud, but I thought it would be useful to consider what it would mean to add Pixel-like machines to the cloud.

First off, I think Google is dreaming here, but it's an intentional act of dreaming as indicated in this article shared by +Tracy L. Crawford. What currently boils down to a $1200–1500 high-end web browser is not a mass market machine. Instead, it's meant to evoke in the minds of people who can do something about it the kind of scenario where the Pixel makes sense. In a nutshell, I think that scenario is as follows:

• Most of your processing power lives elsewhere. Google engineers using the Pixel are compiling on linux workstations back at the office. They're accessing Gmail and internal Google+ hosted by Google.
• You're not hampered by legacy software compatibility needs, in particular the need for seamless Microsoft Office compatibility. It's amazing how embedded office is in the infrastructure of even new phenomena like eBooks or electronic publishing. Microsoft Word is effectively still the standard of exchange for a large part of the content production industry.

Who do I think this most closely fits right now? Students, traditional small businesses not heavily into content production, web-only content creators, and cloud software developers are a few groups who spring to mind.