iSpective

Igor on tech and life

Tesla Autonomy Day

Quite interesting video on today's state of Machine Learning in Tesla. Details on their hardware and software are impressive.

The keynote actually starts at 1:09h:

ML introduction videos

Few nice introductory videos to Google's TensorFlow.

First is about the idea behind ML in general:

Next is about actual problems it's able (and is already) to solve:

WWDC Talks

Here are few nice talks with Apple's top managers right after WWDC:

And last but not least, here's an interesting interview by John Gruber with Craig Federighi and Greg Joswiak discussing the news from WWDC 2019:

WWDC'19 Summary

This was a big one.

The dev conference started with a TV series announcement trailer 😑

WWDC-2019

tvOS 13

  • PS4/Xbox One controller support
  • New wallpapers

watchOS 6

  • New watch faces
  • Hourly beeping sound
  • Audiobooks, voice memos, calculator
  • Independent standalone apps
  • Audio streaming API
  • Own App Store
  • Activity trends
  • Noise app to monitor noise pollution
  • Cycle tracking
  • Health summaries
  • New bands

iOS 13

  • Speed improvements
  • Dark mode
  • Improved Safari, Mail, Notes, Reminders
  • Improved Maps with Street View (SF only as usual?)
  • ‘Just once’ location access option
  • Sign in with Apple - private social login
  • One-time email address generation
  • HomeKit secure video
  • iMessage avatar pictures
  • Enhanced Memoji with earrings, teeth, eyeshades, etc.
  • Memoji stickers
  • Improvements to photo capture and post-production
  • Rotate videos, filters and effects for videos
  • ML to remove photo dupes and choose the best of them
  • Improved Photos.app to remove clutter and let focus on important moments
  • AirPods reading and replying to messages
  • AirPods audio sharing
  • HomePod handoff
  • Siri live radio support with iHeartRadio, TuneIn, etc. 100k stations.
  • Updated CarPlay
  • CarPlay integration with 3rd party apps
  • Neural TTS - much better sounding Siri voice

iPadOS 1?

  • Much improved multitasking
  • Multi-window support for the same app
  • Improved Files.app
  • iCloud Drive folder sharing
  • SMB File sharing
  • Mass Storage Device support
  • External device support via usb-c
  • Desktop class browsing on Safari iPad
  • Download manager
  • Custom fonts
  • Working with text is now much easier
  • Apple Pencil latency down to 9ms from 20ms
  • New notes app
  • PencilKit

New Mac Pro

  • 28-core Xeon CPU
  • 12 DIMM slots
  • PCIe expansions
  • Special dual-core Vega II GPU
  • Custom designed video processing card
  • 1.4kW PSU
  • Optional wheels
  • New 6k, HDR, anti-reflective, 1000 nits display
  • Mac pro starts at $5999, new display at $3999/$4999

MacOS 10.15 Catalina

  • iTunes now broken down into Apple Music, Podcasts and Apple TV apps
  • iPhone sync moves to Finder
  • Podcasts will have content search
  • Apple TV 4k HDR playback on Mac
  • Sidecar - wireless connectivity tablets as second screens/tablets
  • Voice control
  • Improved find my Mac even when it’s offline
  • ScreenTime

Dev Tools

  • iPad apps migration tool to Mac
  • RealityKit - kit for building photorealistic scenes, includes 3d scene tools
  • ARKit3 with real-time people occlusion, motion capture
  • Real-world Minecraft
  • SwiftUI - new UI framework written in Swift - less code, live preview in sim and device (sic!)
  • Native UI framework for watchOS

It was probably the most packed WWDC I've seen. I hope all of the new stuff will be working reliably when it's out 🙂

Support indie developers

Two years ago I supported BetterTouchTool (Mac OS tool for extended keyboard/mouse/touchpad control) with a whooping $5 😃

Better-Touch-Tool

And today after another update out of the very many updates in the last years the developer told that the older license won't work with the latest version and you'll have to buy a new one for $6.50.

I use the app only for one single feature - long time ago I assigned three finger press on the touchpad to simulate middle click which allows me open Safari links in background tabs and close them as well by 'middle-cliking' them everywhere in the tab and not only on the small, hard to target X button.

And even though I use the app for only a small single reason, I bought the new for few simple reasons:

  1. The developer was constantly updating the app through the years
  2. He never even peeped about asking more than my initial $5 for all those updates
  3. He still continues refining and improving the app, which means he still pours a lot of effort into the app
  4. The app saved me thousands of additional clicks for those simple operations I use it for

I constantly hear people complaining about how developers are trying to rip them off. But those people forget that developers are like everyone else - they need food, shelter, provide for their families. And getting $5 in two years is far from what they need to ship a quality product on a constant basis. So even when some of them switch to subscriptions (which I don't like as anyone else) I still understand their reasoning and buy not the product itself but more and more I invest for great people constantly refining them. And I wish more people would understand that as well and appreciate the hard work done to bring us, customers those refined products we use daily.

That's why please support indie devs when that support costs you that famous cup of coffee everyone's comparing it to. Because you will drink that cup and that's it. But the piece of software you buy will make your life easier day after day.

Google I/O '19 Summary

Today was Google's dev conference, and here are the main takeaways.

Google-IO-19

General

  • 'Full coverage' from Google News comes to search
  • Podcasts are coming to search as well
  • AR search results you can see on your phone
  • Google lens shows popular dishes in the restaurant's menu
  • Duplex not only will make reservations over the phone but now over web too
  • Assistant will now have more personalized recommendations
  • Assistant comes to Waze
  • More transparency and control over privacy
  • Incognito mode in Maps, YouTube

Android Q

Innovation:

  • Foldable screen support & continuity
  • 5G support
  • Offline auto-captions for videos
  • Smart Reply for IMs
  • Dark mode

Security & privacy

  • Privacy settings
  • Reminders for location usage
  • Faster security updates without device reboots

Digital well-being

  • Focus mode
  • Extended Parental Controls

‘Helpful home’ devices gathered under the Nest name

  • Easy to use
  • Personal for everyone via Google Assistant
  • High privacy

Smart home:

  • Nest Hub Max - touchscreen device to control your home aka Alexa with screen
  • Previous Nest Hub got a discount

Google Pixel 3a and 3a XL

  • Starts at $399
  • Has a headphone jack
  • AR maps navigation
  • AI enhanced camera with Night Sight
  • Unlimited Google Photos storage
  • 30h stand-by time
  • 3 years of updates

AI

  • For researchers
  • In healthcare
  • To foresee floods

Satisfy the market or leave it

I had the first Pebble watch from a company who was the first to raise $10M on Kickstarter and then the first to raise $20M with their second watch. Few years later this seemingly popular and successful company with all their assets, intellectual property and strong interest first had few layoffs eventually was sold for just $23M to Fitbit. How did that happen?

OnePlus-Pebble.png

It's not because they got fierce competition or were too early to the market - smart watches from big brands weren't there yet and the wearables sector had healthy demand already as well. The problem was that Pebble tried so hard to please their core audience - early adopters that their products never came out from serving that niche to catch the broader market.

Early adopters are a good way of taking your product off the ground: your customers are willing to pay extra to be the first to get your product even if it means dealing with the potential production delay which is common on Kickstarter. But the early adopters are ready to sacrifice their time and money for that new thing on the market and are also much more forgiving to any imperfections and roughness of product's first versions than any regular user would be.

The problem comes after. Since the early adopters is the first and smallest user group in the product's life cycle, holding on to them will never let you move forward and gather necessary resources to do so. After getting your product, early adopters can't wait you to make it better usually for no extra cost. Moreover, each backer thinks his or her new idea/feature should definitely be implemented to the next revision of the product and also retain all previous features as well.

Company's early adopters probably also won't jump into buying the next revision of their product considering the changes aren't significant enough to update and demanding unlimited support for the first version. Also paying a full extra price for something which is similar to what you have (usual iteration of products) especially with similar features is not what early adopters do.

And by sticking to the requests of those first customers, treasuring them companies continue building products which almost no one would buy - for previous customers there won't be much better/newer than what they have, and for the mass market the upcoming products without major changes will be still too finicky, techy and not very appealing. That's why companies which don't adapt to the new users by trying to satisfy the first ones go out from business just not reaching that bigger mass market group.

In this video the author explains the situation in more details with examples that include Pebble's. He also touches OnePlus which for me is the most interesting one and where I myself have proof of the early adopters theory. A friend of mine who is a strong OnePlus supporter is very judgmental about OnePlus' moves in the recent years. For him it's unacceptable for the company leaving out the headphone jack, raising the prices and overall following the industry leaders - everything opposite the brand was famous for when it was listening to their users' opinions back in the day.[1]

But at the same time my friend bought only 2 out of OnePlus' 9 models (ignore the 10th McLaren 6T) literally supporting with money his favorite company only twice! No wonder they had to adapt to the mass market and leave the core loyal fans behind. I bet Apple is successful just because except of pure specs they are selling the image of being an iPhone user so successfully that those users fuel up the company with yearly (2 years max) upgrades. This is where OnePlus is now going and where every other company which overcame the first stage should go to stay in business. Otherwise by sticking to the early adopters you won't be able to ramp up to the next stages.


  1. At some point OnePlus actually had a poll whether to include or not the headphone jack. Unfortunately the poll was ignored and the 6T was released without a headphone jack. ↩︎