Google #io19

Not as groundbreak as io18, but there were still some pleasant announcements in the main keynote. Here are some of my favorites:

Next-Gen Google Assistant

Obv Google Assistant (GA) is already pretty good. But there are an insane amount of deficiencies, limitations, and imperfections. What’s new is that GA can run on-device, locally, without needing to communicate back to online services and all that round-trip audio upload, download latency etc. Super great for most mobile situations in low connection areas, spotty connections, no connection. I’m big on reducing latency. Lots of speed-issues and user impressions is about latency, in everything. And bandwidth in most places are more than sufficient, it’s the latency that’s lacking. Even more so for mobile.

By seriously reducing lag time, running on-device, GA will be increasingly more responsive, fun to use, and actually useful instead of getting in the way.

Next-gen Google Assistant on-device
Continue reading “Google #io19”
Google #io19

LA

Los Angeles, was, unexpected. It was a hard decision thinking/choosing between going to Las Vegas, Los Angeles, or some other small random location in California. Ultimately, the path from LA -> SF seemed more optimal, and Vegas seemed really clishchy (what’s that word).

So, LA. Not really the type of town/city of my style either. LA is characterized by beach, hollywood (film industry), sprawl, cars, hollywood boulevard, entertainment industry, none of which were my thing frankly. So half reluctantly we went to LA – after all, we couldn’t be going to NYC two years in a row (or could we).

Continue reading “LA”
LA

Unifi-ed

My old Ubiquiti Edgerouter Lite died (well it was the second one that died), so I was out of proper routers. Using a spare/emergency Netgear freebie set as a stand-in just didn’t seem to cut it. I couldn’t reconfigure all my DHCP to my liking, I couldn’t get my routing nice, my VLANs, and I suspect it had some scanning turned on that made everything so laggy. Maybe I should have just turned off the scanning.

Continue reading “Unifi-ed”
Unifi-ed

Recommendation engines

https://threader.app/thread/1094359564559044610

https://youtube.googleblog.com/2019/01/continuing-our-work-to-improve.html

Having been on the Internet for much of the past decade and more, it’s been stunning to see how it has evolved. Yes it has, by a lot. The social media platforms, the YouTubes, the other big platforms, have grown and grown and grown into billions of viewers, billions of contributors, and trillions of pieces of content and even more metadata. 

The feel also changed, the feeds changed. It might seem like it was just the mix of people and the mix of content that has changed, but hey perhaps like the american elections and this guy has said, the recommendation engines are partly to blame. 

It’s crazy, isn’t it, that the recommendation algorithms, optimized for innocuous metrics to up engagement, watch time, retention time, repeatability and many more innocent concepts have and seem to promote undesirable material simply because the human mind just seems to love gossip, wild stories, imaginative news and the lure of tabloid content. There must be a reason tabloids and gossip does so well. Hah. It’s dangeous. 

I hope more people are able to distinguish between the truth and the fake, the good and the bad, the real and the unreal, but like our poor sad little human minds, it might be hard to escape the mental clutches of logical fallacies, dreams vs realities. 

Recommendation engines