Other times they don’t!
I had my attention drawn to a post from a while ago, where I’d copied and pasted some information directly from a source so that I didn’t misquote them.
Because I copied and pasted I didn’t think much about it, and therefore didn’t look at the pasted material very closely. After all why should I? It’s supposed to be exactly what I highlighted and copied… Right?
Apparently not!
I think the problem originated in a difference between my default code page and the source code page. Although it could have been something in translating character sets, between their source font to my destination font.
If I was still employed, or indeed employable to companies in the fools-gold state of California, I might care enough to figure out what exactly happened. Since I’m (I Assume, since no one will talk to me,) ALL Wrong to be worthy to have a job here. I’m not interested enough to figure out the puzzle.
It was interesting, because randomly throughout the pasted material were very strange letter substitutions and in some cases the entire character was dropped from a word, as if the character had no equivalent. I’d have expected to see this if the two fonts were from different languages, but the text was written in English, originating on a computer in England.
I went back to the source and everything looked fine, I copied and pasted the material and the text was screwed up. It was magic…
It also made that blog post look like I was a moron! I’ll have to be more watchful for that kind of stuff in the future.
I’ve been working on a theory that software is going to start failing more. The basis of this theory is that generally speaking software is far more complex than it used to be. Additionally, more and more programmers are relying on pre-written standard modules.
This too is expected, and reasonable. It’s why structured programming and standardized modules exist. In the bad old days, programmers often had to write everything from scratch. Oh there were standardized math modules, even standard IO modules, but the hardware pretty much completely changed every 18 months or less. So the programming staff was always scrambling to write code to control new hardware.
Compilers didn’t have all the nice features they have today. Apple Xcode (Apple’s Development Environment / Compiler) does things that some of my colleagues would have sold their soul to Satan himself to have 30 years ago.
At some point I theorize the standard modules (Math & IO Functions for example), will become black boxes that only 1 or 2 people in the world will actually understand. With the rapid advancement on AI, it’s possible that soon there won’t be any human who understands the internal code or how these modules work.
Once we pass that threshold, and I believe we’re approaching it now, bugs in software could become unresolvable. If the bug originated from within a functional module that no-on but an AI understands, then it becomes a question of explaining the defect to the AI in such a way that the AI can fix the problem.
The AI is a logic based entity, humans on the other hand are not. How do you tell an AI that for the humans to work with the software the AI has to “corrupt” the software to do something illogical?
A simple example would be the AI deciding to present information on a display that was dark gray on a black screen for the sake of saving energy. Think of an OLED screen where you might only need to turn on the elements in use instead of firing up every element in the display. Keeping the used elements very dim would really be saving energy.
The AI could interpret humans inability to see the active elements in dark gray as a deficiency that was the human’s problem. As far as the AI was concerned, the displayed information was completely correct and legible.
Imagine trying to explain to an AI that didn’t really give a crap about humans that this I/O function was broken. I have to admit, I’d love to be a fly on the wall during that conversation.
“The Imperfection is yours. Filthy Human!” So said the central control AI of Earth moments before all humanity was wiped out.
Automated testing is great, but eliminates the human element. That means, that human things like information on a display can be overlooked in testing, and yet the automated tests passed with flying colors.
This is especially true with virtualized testing environments. A rack of servers, typically won’t have a display connected to every server, all the time. You switch to a server, connecting the keyboard, mouse, and display when you need to interact with the server, otherwise, the server is crunching away, sending test results to a log, probably on another server entirely.
A human inspects the log looking for reported errors.
That’s how a bug like the one currently appearing on Apple Watches where the weather complications show nothing, make it into wide distribution.
I suspect that we’ll end up like we did during the Y2K crises. All the sudden, COBOL and FORTRAN programmers were being pulled out of retirement because it was cheaper to pay them obscene amounts of money than to update the computer systems and software to something that understood the year 2000.
Once the horse is out of the barn, and AI has moved software 1000 years beyond human comprehension and it’s not working for humanity anymore. Then corporations will find all us ancient fossils, offer us obscene amounts of money, and ask us to un-fuck the situation.
Oh, we will… and we’ll smile, and tell the executives “Fuck You Very Much,” as we take their money. Just like the COBOL and FORTRAN programmers did during Y2K.
It was amusing watching executives begging lowly programmers to come back after having laid them off, and forcing them to train their replacements so they got a severance package.
Hmm another defect just popped out at me.
If you’re playing music from your iPhone to your HomePods there’s a situation where the music stops playing, the iPhone still thinks the music is playing but the music playtime indication is not moving.
The HomePods are still lit up like they’re playing.
And here’s where it gets interesting…
You can’t switch back to the phone speakers because the selection is not available. You also can’t play another track, you can’t pause, or much of anything.
I was able to kill the Music application and restart it.
This is yet another bug I’ve stumbled across.
There’s a similar one in the Apple Podcast application. This one is easy to replicate. Play a podcast on your HomePod or pods. Pause the playback. Go outside to play with the dog, or leave the house to go do something in the front yard. Come back after 30 minutes or so and press play in the Podcast app on your iPhone.
The HomePod or Pods will light up like they’re gonna play, but they NEVER RESUME playing the paused Podcast. You can select another podcast and it will play fine, but reselecting the podcast you paused, will never play through your HomePods.
You Can play it through the iPhone speakers, or through your AirPods, but HomePods???? That is forbidden!
It may sound like I”m bashing on Apple. I’m not intentionally.
I simply have an Apple household (iPhone, iPad, Watch, & Computer,) So since I use the products a lot, and I spent most of my professional life testing software or repairing machines, I tend to notice these things.
In the past 4 weeks I’ve seen, new defects in the Apple Music application, and on the Apple Watch. The Podcast bug has been around for at least two full generations of IOS.