This whole blog is just a regurgitation of surface-level wikipedia knowledge. All of these issues are very well known in the industry. Asking to solve all of them, perfectly, at once, is clear indication of someone who's never done product development.
Yeah, these kind of people are the worst in any kind of product ideation session. Nothing is ever worth doing to them because all they do is point out minute flaws the whole time. Contrarianism taken to a level beyond useful devils advocate.
The other obnoxious thing here is that he thinks he’s being novel? Like Apple hasn’t considered any of these points over this headset’s near decade-long gestation period?
Ultimately though I’ll at least give him credit for putting himself out here so publicly. Plenty of people are harboring private doubts, but he’s setting himself up here for potential posterity, up there with all the people who said the original iPhone would be a failure.
I believe the author has an extensive background in product development in the area, which is why I was surprised that GGP thought they'd never done it before.
Gigapixel is no longer in active development, they’re focusing on Topaz Photo AI (which rolls all of their AI photo tools into one app, including upscaling) and Video AI. Both are pretty decent considering they’re just recently out of beta, and they do roll out updates with fixes and new features fairly frequently. I’d say they’re both pretty good at what they do but on the photo side you can definitely get some decent results using free tools and models if you have the time to play around.
A lot of these large companies spend millions of dollars recruiting participants from various demographics to test prototypes and fine-tune technologies like this. It's a tremendous effort.
A lot of these large companies also, for some undisclosed reason or chain of events, design a phone so that a non-neglible amount of people can't hold it without dropping calls (iphone 4), and then blame the user for "holding it wrong".
Also such a thing as being too conservative with battery size (iphone 6?) so that when it unavoidably degrades, you must throttle down the phone to avoid killing the battery and perhaps more (literally by a current draw it can't handle since the internal resistance has increased).
And that's just apple, then there's unintended acceleration (toyota), exploding airbags, faulty ignition (ford was it?).
Don't think that just because they are large, or a company, or old, or whatever, they always have their shit together. Companies are people. People make mistakes. Mistakes can slip through the cracks.
I believe you, and it makes sense. Apart from that, though, it's hilarious to imagine an ad like "Wanted: Black women with long hair to quietly walk into a room, pick up a vase, drink a can of coke, and then leave through the same door."
Yes. Although it's damn hard to recruit a Spanish stay at home mom with a slightly twitching left eye who's too busy to participate to random studies and will receive this a gift for her 50th birthday.
It kinda makes sense that some of her traits would be covered inside the test pannel, but looking at the face unlock and the level of adjustments they're still doing, it doesn't look like they got nearly enough data before launch. And that's for a product that is selling in billions of units.
I had similar thoughts for a while. What helped me was instead of aiming to make all of the time outside work productive (3-4 hours) just aim for under an hour. For me, 40 min of producitivity per day outside of work makes me feel accomplished. You can actually get a lot done in that span of time like do some exercise, practice an instrument, read an article... (pick one per day and cycle through the week). Not much immediate gratification, but over the years it adds up to a tonn of personal development. Plus with a 40-min window you don't feel guilt about abandoning your other family duties etc.
What's interesting is that the trend in "Average Weighted Accuracy" seems to increase linearly in the past three years whereas the number of model parameters and required compute is growing exponentially. So accuracy is pretty expensive.
I mean, yes, you have around 100 trillion interconnections in your brain, likely the most complicated biological device on this planet, to create the accuracy you perform. We're just very power efficient when doing that computation.
Same is unfortunately true for a lot of Eastern European music. There are thousands of great soviet recordings of various classical and folk pieces that have never been digitized. And the current atmosphere over there is not honed in on preserving art. A lot of the recordings are only available on 50-60 year old vinyls which are getting harder and harder to find in a decent condition. Been digitizing a lot of my favorites in my spare time.
