Post has attachment
Gerwin's take on I/O

My thoughts will be coming later this week after I catch up on work and some of the announcements I missed, but I spent a lot of time with +Gerwin Sturm during the conference festival, and his thoughts are a pretty good reflection of mine.

(And I love this picture he took of the I/O sculpture. I have a picture of him taking it, and its not nearly as good.)

Post has shared content
Checking out +Firebase over the last couple of days has been really cool. The biggest problem is that web is currently a second class citizen behind Android and iOS. It doesn't get access to analytics, crash reporting, notifications (display and triggering APIs), and offline databases.

Hopefully those will come to web soon as the real power of Firebase is everything working together. Plus I want to build real progressive web apps!


Post has attachment
Some thoughts on +Sundar Pichai

I first saw Sundar back at I/O 2012. He was the main speaker on the Day 2 keynote, and he talked about Chrome and Chrome OS... and said we'd get a ChromeBox.

Since then... his star has definitely risen. I/O 2014 was his show from top to bottom, as was his passing reference to the Cardboard we'd be getting on our way out the door. A what?

And here we are at I/O 2016. CEO of Google. In charge of it all. The buck stops here.

More than once, when I challenged engineers or DevRel folks about not being able to trust Google to not screw things up, their reaction was similar: "Sundar is in charge now." Can one man really change the culture at Google? Really make people trust Google again? I dunno... Firebase and Spaces and Allo/Duo make me wonder. But those are old projects released in a new regime. Perhaps he can make them not play stupidly together.

I wish I knew what he really thought. What really drives him. You can hear him talk about AI and ML, and he becomes more animated. But can he lead a company based on that? It is difficult to tell, but I think he has the drive and the vision to pull it off.

I hope so, anyway.

Post has attachment
Rare Breed indeed

Lots of press prowling around for interesting stories at IO. +Jillian D'Onfro​ was interested in those of us still using Glass for years after it was literally launched at IO. A few of us expressed how it is still our go-to technology.

Post has attachment
IO Day 3

Winding down.
Last questions.


And, we hope, eventual return.

Some more thoughts to come, but a bit of rest and reflection first.

Post has attachment
McDermott Report: Live from I/O

I'll be on the air with +Dan McDermott in a few minutes to talk about Google I/O. Join us! Comment! Convince me not to strangle Dan!

Post has attachment
From the Source of Spaces

I chatted a bit with the Spaces team at I/O. They were standing next to a big magnetic board where people could write up their "Big Ideas" on a tile, create a space, and then have the invite URL for that space set to the NFC on that tile. It was a very clever way to onboard people into a space.

I've had almost no traction on that space. Nor on many of the spaces setup for each session.

I asked them about how they expected people to normally be invited to a space. "Email or social." "Why not use that forum to then carry out the conversation?" "They're not good at threading." I challenged them on that point, since GMail has had threading for... what... 10 years? "You can't look at an item and discuss it at the same time."

Is this really a problem? Are Spaces actually solving any problem anyone has?

I raised things like being able to have useful things in Spaces, like a shared calendar. "This is an early release. We're still experimenting and looking into it." Integration with Communities. "We're looking into it, but we want to see how it works stand-alone." Discovery and an index of public Spaces. "We are worried that will create unfocused conversation." Pinned posts and everything else we have in Communities and Groups. "This is an early release, and we're experimenting."

I'm suddenly tempted to go back and ask them if they really have data showing that searching for stuff inside a social space was what people wanted and would find useful. As opposed to all of the stuff I asked about.

API? "This is an early release, but we don't have plans for one." They didn't even ask me how I would use an API. What I would want in an API. What my use cases were. What the story would be for Spaces with an API. Nothing. Total lack of interest.

There are a lot of things I think are cool about the concept of Spaces... but I don't think they hit the right notes with this implementation. I think it shows a lack of understanding of even the use cases they envision.


Post has attachment
Google Atmosphere

Last minute addition to the program. In light of the heat on Wednesday and the cool wind today, the ATAP team, in conjunction with the Loon team, demonstrated the latest project... Google Atmosphere. Based on their Google Brain models, they can now predict microclimates and change them.

Some limits. Can't make it suddenly snow, for example.

Public availability this Fall.

Post has attachment
My Firebase Epiphany

I've spent a lot of time grilling various Firebase people about... what this new thing is. From a year getting familiar with Firebase as a great all-platform real-time database... it has suddenly become... what?

Well, the "what" is essentially a mobile library for Android and iOS to provide access to a wide rage of services. That isn't the epiphany. They made that really clear thats what it was now.

There are lot of things missing right now. I asked, on behalf of +Linda Lawton, if there was API access to their console, particularly Analytics. Their answer was that there was access where it made sense... but Analytics wasn't yet available directly. You can dump everything to BigTable and do queries there... and you can assign things to tags, and it will go to both Firebase Analytics and Google Analytics. But right now there is very little crossover. Their line was that Firebase Analytics were aimed at mobile... and Google Analytics was aimed at the web... and the needs were different... blah blah blah. They didn't seem to understand that people needed to support both, so fragmenting their platform like this would make our lives difficult. They suggested we complain loudly on their support forum.

I asked a lot of times about why this branding. They all repeated, as if from a script, that the Firebase brand had a focus and reputation on their clients. The cynic in me said "you mean unlike Google?" I think their argument is still weak... but that also wasn't the epiphany.

I need some Android folk to back me up on this, but I'm wondering if they're doing this because of the problems they've had with Google Play Services. GPS is tied to Google Play, which means it is tied to partners who have licensed Google Play. Which means that you can't have Google apps on the Kindle or on a wide number of Chinese phones.

What if you had a library that wasn't tied to Google Play? You could have people use Google Ad Services without having to license Google Play! That appeals to people writing apps so they run on Kindle... it appeals to people writing apps for the Chinese market... and that means it appeals to Wall Street who were wondering what was going to happen to Google's ad revenue.

I think we will see this as a trend. Moving stuff out of Play Services so those services can be available on other platforms. That brings people into the Google ecosystem, and Google's various revenue streams, without forcing them onto a particular place to purchase the app.

Nobody I've talked to has said this was the reason. I haven't seen anyone else come to this conclusion. It may not be the reason. But I think it is no surprise that some of the core elements of this Firebase library have to do with market segment targeting, targeted delivery, analytics, and advertising.

It is slick, clever, and very very smart.
Well... everything but the name is, anyway.


Post has attachment
machine learning: Google's vision

Because of mobile, real world problems are now AI issues. Can you scale your UI to match the AI?

Broad set of models will be available to those less familiar with building those models. TensorFlow tutorials are helping people understand this.
And for those even less familiar with ML, there are and will be libraries to just use.

Very hard to make broad domain conversational AI, but easier for nanodomains.

"The best UI is no UI"
"AI becomes ambient across all your devices"

Wait while more posts are being loaded