Google [X], Part III

Here’s me, a year ago, sounding crazy:

I bet the Google self-driving car never becomes the market leader. It’s possible it’s not even the first to market for public purchase. It’s even possible that it never makes the market at all.

I bet Google rolls back into a software partner role in its alliance with Ford. I bet Google distributes its software for cars the same way it distributes its software for phones.

And here’s the news this month:
Google has refocused its autonomous driving efforts away from developing its own car without a steering wheel and pedals … Instead, the company is said to be partnering with automakers on more conventional cars
Shortly after that news broke, Google spun its car division out into Waymo, which still sorta promises to make self-driving a reality. Turns out I was a little wrong — it may wind up being Chrysler or Honda who partners with Google, not Ford. But I’ll still call the harder prediction, that Google won’t be market leader or even first to market, a success so far.
It turns out cars are pretty hard to make.

Don’t Vote, Part II

In Don’t Vote, I wrote about how you could multiply your voting power by not just voting, but campaigning to get others to vote — and that thanks to modern technology, it’s now cheaper and safer than ever to campaign.

Scott Adams, a better writer (debateable) and a bigger asshole (certain) than me, points out that the reverse is also true: You can also convince people to not vote. Ten extra votes for your candidate of choice and ten missing votes for the opposition are about the same currency.

(PS: if you’ve somehow registered to vote but are as of yet undecided on whether you’re actually going to do it or who you’re going to do it for, please vote for hillary, the election is still closer than it seems, and your vote matters. thanks)

Google [X], Part II

I wrote here about how I thought Google [X] isn’t really a research company, but subterfuge employed by Google for entirely different ends.

John Gruber’s Daring Fireball links to a pair of articles which add some supporting color. The focus is on [X]’s reported 2nd quarter loss of $859 million which at first sounds kind of bewildering… but check out the far more interesting quote near the end of his wrap (emphasis mine):

The combination of big ideas, lofty rhetoric and a strict code of secrecy has made X a source of endless speculation and conspiracy theories. The one you hear most frequently, usually from competitors and venture capitalists, is that X is a giant public relations plan to distract regulators from Google’s search business, which is under scrutiny around the world. 

Not precisely the ruse I had in mind (to be fair I don’t often run in circles of VCs), but also not incompatible with the theory I proposed.

Worth illustrating:

Google executes 60.4% of all desktop searches. On the Herfindal-Hershman Index, which measures market concentration, an industry score below 1,000 indicates adequate competition and a score above 1,800 means you should expect scrutiny from the Justice Department.

… The HHI score in desktop search: Juuuust a tad over 4,700.

So knowing that Google made $20.1B revenue in Q2 2016 in a search advertising industry that is screaming for antitrust intervention… I’d say that an $859mm loss (just 4.3% of revenues) is a pretty justifiable expense to keep regulators off your tail.

Best Buy & Amazon, Part II

In September 2012, I wrote about how Best Buy had the potential to beat Amazon in the home delivery game, specifically suggesting that Best Buy might leverage its storefronts as a network of warehouses, and further suggesting that one-hour delivery and fulfillment of online transactions might be within reason.

Two things have happened in 2014.

First, Best Buy has seen a marked improvement in company performance, which investors attribute to Best Buy’s edge in using its storefronts as shipping hubs.

Second, Amazon has launched one-hour delivery in NYC, with plans to add additional cities in 2015.

I don’t think I was even really serious about the one hour window thing back when I wrote the original post. So it’ll be all the more incredible to see how that feels in reality the next time I have to buy Amazon junk.


Netflix, Part III

In January 2011 (almost four years ago!) I wrote about how Netflix should let me recommend movies to my friends.

In fall 2014, Netflix came out with precisely that feature.

I actually had a chance to chat with someone on Netflix’s Product team sometime around February 2013, who suggested that they’d been kicking the tires on the idea for some time — with one of the biggest hurdles being that they wanted to be conscious of users who really didn’t want to share their viewing history openly.

I’m really glad that the feature is out in the wild now, and that the team pays such careful detail to its users’ privacy. I’ll still be looking forward to my  Jan 2011-proposed combined-profile algorithmic recommendations, so that I can easily figure out what to watch with my parents whenever I’m home for the holidays… hopefully we’ll see that before another four years pass.



Cars & Smart Screens, Part III

In March, I wrote about how cars should solve their most glaring dumb interface problem by abandoning the mirror system that’s been in use for over a century and upgrading to modern camera/video technology.

This month, Cadillac revealed that a video feed rearview will be a standard option in one of its lines in 2016, noting that this solution will “offer a field of view up to four times greater than that of a standard mirror,” and also “reduces glare and allows crisper image in low-light situations.”


Spotify, Part II

Spotify announced today that they’ll be launching a free playlist-based product for mobile devices, with an eye towards increasing user trial and adoption.

For smartphones, you can access all of your precompiled playlists, as well as use the Shuffle product. However, you can not perform unlimited search and listen queries.

About a year ago, I wrote a post about Spotify Mobile Playlists which called for very nearly the same thing:

Can Spotify have its advertisers curate (and pay for) playlists, which I can listen to or even temporarily download for free for a period of time?

Let’s say Pitchfork, or heck, even Harley Davidson, now let me download a playlist which I could listen to for a week. Instead of ads after every other song, build in liner notes. … Bits are catered so that they’re both relevant to the sponsor and to the music, so that all of a sudden the ads aren’t an apologetic interruption to the listening experience.

Chalk another one up to the “I love it when I’m right” column.


Re: Netflix

Two and a half years ago, I wrote what was probably my most efficient post ever (in terms of quality-to-wordcount ratio). Netflix, I claimed in 90 words, was a really obvious destination for relatively basic social integration, with use cases for sharing that already mirror or enhance real life media-consuming behavior.

I’m amazed that “Netflix Friends,” and being able to find mutually interesting movies, haven’t come to pass yet.

Instead, the company recently unveiled Netflix Profiles. On the plus side, I guess this means the product team has a pulse. But really, if you ask me, this idea is almost entirely backwards.

Per this new feature, you can now add up to five different profiles to a single Netflix account so that the movies one “member of your household” watches don’t get conflated with the recommendations for another member. That’s fantasy. In reality: The “members” of my current “household” are me and Kevin, my former roommate who presently lives in a different timezone. I wouldn’t be surprised if our family even had a handful of other “members.” With Profiles, it’s now even less painful for me to leech off Kevin’s account. One of the very few reasons I’d ever consider starting my own account (aside from the feeble complaints aired by my moral compass) was for the benefit of having my own personal recommendations surface.

Why would Netflix create a free feature that works as a direct substitute to the company’s primary source of revenue? Isn’t that completely insane?

Furthermore: Almost all of the imagery Netflix uses looks like this:

Fetflix Namily

A family, or group, sitting around to watch something together. That seems to be the primary, ideal use case. Why would you build a feature designed to support doing the exact opposite?

If they’d had their hearts set on multiple profiles per account, then fine. What they should have done, though, is set up the mutual recommendation algorithms first. Not just because this is far more in-line with the primary depicted use case. But also because once that’s in place, there becomes much more concrete, discernible value in having five individual profiles — you’d thus be empowered to surface great, discrete movie suggestions for the entire family, or for just Mom & Dad, or for just Dad & son, and so on. That’s really powerful stuff. And instead of being a feature that’s given away for free, I’d argue that you’re now sitting on a paid premium feature which might even drive an extra $1-$10 (or, roughly, an extra 10%-120%) per month per account to your bottom line revenue.