Disqus - Latest Comments for alistairc

Re: Canadian Startup Entrepreneurs Need To Be Cocky

alistairc — Thu, 10 Dec 2015 06:18:17 -0000

LOL. Great minds and all that.

Re: Redlining for the 21st Century

alistairc — Sun, 09 Mar 2014 22:22:11 -0000

Couldn't agree more. I like your distinction about choice versus control.

I wrote a bit about this new redlining in 2012, and the follow-on from folks like the Oxford Ethics department was fascinating. At the time, some readers chided me for calling it a civil rights issue—but two years later, that criticism seems to have vanished.

http://solveforinteresting....

Re: Minecraft for Education

alistairc — Thu, 09 Jan 2014 10:13:58 -0000

Totally agree; can't wait until my daughter is old enough for this. I also think Civilization is a great way to teach world history, done right.

There are a bunch of interesting languages (MIT Scratch is one) aimed at coding too; and we got a Robot Turtles set from Kickstarter, which is basically Logo The Board Game.

I lament the reversal of computing. An Apple ][ used to sit there, blinking patiently, and you could as easily type LOAD and launch a game as you could type LET A = 10 and start writing code. Today, despite the fact that computers are branded with slogans like "what do you want to do today," the answer is clear: the computer, and the network to which it is connected, will tell you. Anything that restores that balance is great.

Re: A Better Way to Tackle All That Data

alistairc — Tue, 13 Aug 2013 21:12:13 -0000

Good post, Chris. At a recent weekend event it dawned on me that two things I'd been in a lather about—Augmented Reality and the Internet of Things—are really two sides of the same coin. The data created by the IOT overlays the information we perceive through AR. One without the other is either information overload or pretty pictures.

Similarly, Big Data and machine learning are two complementary technologies. Making sense of reams of data means exploring, and assisted exploration is the only way to do it well. I do remain concerned, however, about the widening of the digital gulf that this creates.

It's no secret that there's a lot more data today. Not just voice, video, and text; but an abundance of signals from our cell phones, our cars, and yes—even our thermostats. This data alone hasn't much power. But marrying it to unlimited queries and self-tuning algorithms creates a major problem.

Consider your fourth amendment right: " to be informed of the nature and cause of the accusation". This is widely interpreted as the right to see the evidence against you, and is the basis for discovery in legal proceedings. When someone else has not only the evidence, but also machine learning, unlimited queries, and more, you lose your rights.

I suspect we will see the fourth amendment interpreted differently in the future: "You have the right to see the evidence against you using the tools of your accuser." If this is not the case, we'll have lost a significant chunk of our civil liberties.

Re: Announcing Kiltr Ventures

alistairc — Mon, 01 Apr 2013 13:43:06 -0000

Seems like an auspicious day for a change. Ben and I decided we gained so much weight writing in the last year we're going to write a book on using data to build a better waistline faster. We're thinking of calling it Lean Analytics.

Re: SXSW, bikes, and the Zen of finding things out

alistairc — Sun, 31 Mar 2013 16:24:54 -0000

Wow. Thanks for sharing what sounds like a rollercoaster of emotions. I guess I felt that by treating everything as a learning experience, I was able to let go somehow—everything wasn't bad or good, it just was. Not half-empty or half-full, just equal parts of water and air. ;-)

Re: Lean Analytics

alistairc — Sun, 24 Mar 2013 12:32:01 -0000

Sorry if this gives that impression. I've been a big fan of balanced scorecards since I first came across them in university. Since that time I've used them in sales and product management, but always well beyond product/market fit and with departments, not the company as a whole. They work well enough, but (in my experience) quickly bloat to become somewhat dogmatic job descriptions.

Applying Lean thinking to the model works better. In some ways, the middle three steps of our framework (stickiness, virality, revenue) built on Eric's three Engines of Growth are akin to the "performance hubs" of scorecards. But we find that in earlier-stage companies, or environments characterized by high uncertainty and rapid change, two things matter more than the completeness of a scorecard: focus and resiliency.

Focus can be achieved by picking one thing to optimize at a time. Indeed, when software architects try to optimize code, they do so in exactly this way. It's the only method by which to determine causality (for example, which aspect of the business today is a causal leading indicator of the business tomorrow, so we can change the future by tweaking the present.)

Resiliency comes from narrowing the span of control. I once spoke to a member of the Israeli defence ministry who told me that the Hebrew word for "four" and for "many" sounded similar, and that they often thought of counting as "one, two, three, many." We were discussing span of control, and how many organizations favor wide span of control (and processes) in order to scale, rather than narrower organizational structures with less direct reports. "But in Israel we're optimizing for resiliency, not span," he explained. "To us, three is a good number because each node in an organization can know what its reports' reports are doing—so the organization tolerates damage to the structure better."

What we've tried to do is present a framework for continuous learning, tied to specific gates, that emphasizes focus and resiliency over completeness and scale. I don't think we've ignored recent developments that have applied Lean concepts to scorecards; in fact, I hope we're in strong alignment with those developments. But if you have content you think we should review, I'd love to learn from it and include it in future editions of the book.

Re: The business singularity

alistairc — Fri, 01 Feb 2013 20:33:41 -0000

I butchered math and physics for the sake of readability but didn't see any geography..

Re: The business singularity

alistairc — Fri, 01 Feb 2013 09:22:15 -0000

Yeah, oversimplification is the thief of accuracy. My first draft had a lot more math and put me to sleep. Hopefully the intent still holds.

Re: Using Data Wisely

alistairc — Wed, 22 Aug 2012 16:01:52 -0000

Great points. I find it particularly interesting that optimization may be prejudiced without meaning to be. For example, there are bank trading algorithms in public banks that can't be used because they were created by evolutionary algorithmic optimization; regulators require that banks be able to explain their algorithms, so black boxes need not apply.
But private traders can use whatever they like, creating a two-tiered model. One can imagine that web optimization algorithms notice a pattern (IOS browsers pay more money) and optimize for that. There's no malice—just a pattern.

Re: Big data is our generation’s civil rights issue, and we don’t know it

alistairc — Wed, 22 Aug 2012 15:44:03 -0000

I was actually referring to the compute time—if I have to run a query that involves lookups on an external system, they'll be slow. As the late Jim Gray pointed out, compared to the cost of bandwidth, everything else is free.

But slowness in terms of reflection and taking the time to understand things has to be weighed against the need to fix serious, pervasive problems with the world around us. For example, fraud detection isn't easy in Europe, in part because of privacy legislation.

Re: Big data is our generation’s civil rights issue, and we don’t know it

alistairc — Wed, 22 Aug 2012 15:42:01 -0000

As I understand it, GINA (the Genetic Information Nondiscrimination Act) would protect Californians specifically around healthcare. Personally, with the EPA and other agencies being torn down, I'm not sure I'd put my faith in legislation entirely.

The broader issue is that in a capitalist, free-market world, you aren't being harmed or prejudiced against; you're being given the opportunity for premium pricing or preferential treatment. This tragedy of the commons problem is systemic of public/private blended services like healthcare (as a Canadian, I see this all too well.)

Re: Three kinds of big data

alistairc — Wed, 22 Aug 2012 15:29:12 -0000

Chris,

Someone else pointed out that Healthcare was a Big Data use case. I guess as a Canadian I lump civil defence and healthcare into things the government handles; maybe that's my own myopia.

Good point.

Re: Three kinds of big data

alistairc — Wed, 22 Aug 2012 15:27:47 -0000

Point taken.

Re: Civil rights implications of Big Data

alistairc — Wed, 15 Aug 2012 07:43:50 -0000

Crowley, but damned close. ;-)

Re: Civil rights implications of Big Data

alistairc — Wed, 15 Aug 2012 07:43:11 -0000

Interesting—"privacy isn't what they know, it's how they act on it" is certainly true in this world. It's like in the village millennia ago: everyone knew you, and likely your secrets, through the thin walls of the hut. But if they didn't act differently, it didn't matter.

I was listening to NPR yesterday in Boston, and they were talking about Logan Airport's screening practices, and involuntary discrimination. It's something innate in humans to behave differently if we have subconscious biases.

But as Jonathan Haidt so eloquently explains in The Righteous Mind, our conscious brain is like a lawyer for our moral reasoning, grabbing hold of any cue or clue to defend our reactions after the fact. The issue here is that Big Data might give that internal lawyer a whole bunch of "case law"—seemingly just, reasonable, scientific explanations that are based on predictions about a person, but aren't accurate.

Re: GoInstant To Be Acquired by Salesforce

alistairc — Tue, 10 Jul 2012 09:54:01 -0000

So happy for you guys. And you really deserve it too. I remember saying to you "you're moving to Halifax to do WHAT?" only a few months ago. Nicely done.

Re: Can Big Data Replace Domain Expertise?

alistairc — Mon, 05 Mar 2012 15:08:38 -0000

That was Jeremy Howard, of Kaggle. It was a pretty interesting discussion; there are good points on both sides and the point of the debate was to polarize that. It seems to have worked.

The data proponents point to many competitions in which data beats experience (picture Moneyball, with a roomful of domain-expert scouts nay-saying the new data expert.) Of course, understanding which data to apply to a problem, and when to listen to the numbers, is a nuanced thing.

One thing about data is that it often has non-obvious, and disruptive, nuggets within it that threaten the status quo. And many "domain experts" thrive on their political skills rather than their actual results. So part of the debate is really about housecleaning to replace anecdote with evidence—an uncomfortable cultural shift.

Re: Who Owns Your Data?

alistairc — Wed, 12 Jan 2011 17:22:25 -0000

Michael,

Completely agree with you on the net neutrality example. I think data neutrality is something we won't realize we want until it's too late -- much as we're seeing variable rate-shaping and the exclusion of wireless transmissions from neutrality. As with that argument, bad language and intentional obfuscation get in the way of a real discussion.

Re: Who Owns Your Data?

alistairc — Wed, 12 Jan 2011 16:20:29 -0000

Thanks, Bruno. To be fair, I think I heard someone say these words at a Strata planning event in New York, so I can't claim to have thought of the example myself.

Re: Who Owns Your Data?

alistairc — Wed, 12 Jan 2011 16:19:40 -0000

Thanks for the comments. Shaun, I agree that the "means of analysis" harkens back -- but there wasn't an open source cotton gin, or an open source printing press. The nice thing about bits is they're free to run, so a dedicated bunch of citizens can fire up a cloud and analyze too, whereas a dedicated bunch of citizens couldn't manifest a steam engine from thin air no matter how well-motivated they were.

That's a big (and to me, optimistic) difference between the Industrial Revolution and Industrial Data. The catch is, with net neutrality, regulation, and vague guidelines about what can be done where (see Amazon's handling of Wikileaks' servers, for example, which makes it clear that you run a cloud app at the provider's pleasure.)

So while I agree that everyone needs privacy literacy (and thanks for the hashtag; hadn't seen that!) I do feel that ubicomp and utility computing has made it possible for a reasonable cross-section of citizens to do analysis. Not sure that's enough to save us, though; even doing the analysis generates its own data for corporations to feast upon.

Re: Is reading your wife’s email a crime?

alistairc — Fri, 07 Jan 2011 14:17:48 -0000

And the reverse is also true: your spouse's search history may be grounds for a conviction!

http://arstechnica.com/tech...

Re: Why Is Reddit All Over Digg Right Now?

alistairc — Mon, 30 Aug 2010 11:53:26 -0000

The new homepage is just approved sources. Reddit was part of the alpha program for this. Digg autosubmits all of the articles they pull from an approved source's feed.

Therefore, Digg is pulling in content from Reddit automatically as a result of their alpha program.

Sorry to disappoint the conspiracy theorists.

Re: Lean analytics: Questions VCs should ask (and you’d better answer)

alistairc — Thu, 17 Dec 2009 08:54:02 -0000

Absolutely agree, but this was about website metrics specifically. That said, Kampyle and others do a good job of soliciting feedback with things like "rate this page", which can be overlaid onto other metrics. There's a whole world of call center response metrics that can be tracked; if your business has a significant customer support aspect to it, then that should be a separate part of the board meeting (with the owner of that department reporting.) Linking the two together to correlate, for example, long call hold times with reduced revenue, is essential.

Re: Google Analytics Alerts: the start of a complete view?

alistairc — Fri, 27 Nov 2009 09:22:33 -0000

Eran,

You'll notice we mentioned Kampyle in the post above. We also had the folks at Kampyle review Complete Web Monitoring, and cover both the Greasemonkey and API integration of their product with Google Analytics. We're big fans of client-side aggregation techniques that pull multiple dashboards into a single view!