dewme

About

Username
dewme
Joined
Visits
930
Last Active
Roles
member
Points
15,786
Badges
2
Posts
6,108
  • Apple confirms that Apple Intelligence Siri features are taking longer than expected

    charlesn said:
    CNBC is reporting that the Siri improvements are now delayed til 2026. That's the headline of the article currently running on the website. This (obviously) would mean an entire year of additional delay beyond a launch date that was already pushed out nine months from when A.I. Siri was first announced. My "Apple memory" goes back to roughly 1990 and I can't ever recall something like this happening before--a product announced for debut nine months later (that in itself is rare enough) and then blowing by that debut date by a full year. So I guess AI Siri now debuts in time for its sweet sixteen--but you really have to wonder if Apple is ever going to be able to fix Siri. smh. $166 billon in cash on hand and we can't do this? For context: just Apple's cash, if it were a separate company, would rank #90 on the list of the 100 most valuable companies.  
    I know that nobody wants to hear it, but developing software is incredibly difficult. If you look at the way hardware has been scaling up over the past decade alone, it's absolutely phenomenal. I'm not implying that hardware is not extremely complex, it is anything but, especially with increased specialization with CPUs, GPUs, TPUs, NPUs, cache logic, etc. In these areas the complexity has increased in each functional domain, but the numbers shoot up massively as a result of parallelism. 

    Once a specialized hardware design is built out in one of these areas, optimized, verified and validated, it is repeated numerous times over to achieve greater parallelism and multiprocessing capability and capacity that scales out as far as the fabrication technology and physics allow it to.  The M3 Ultra's 180+ billion transistors attest to the massive scaling that is possible. But once it's certified as being done, it's done. 

    Software at all levels including, machine level, microcode, firmware, drivers, kernel, system, application, etc.,  are all built to take advantage of the capabilities provided by the software layers below them and ultimately the hardware.  The software layers can be changed tremendously over the life of the underlying hardware. 

    At each software layer there are humans involved in generating the requirements, designs, carrying out the implementation, testing every function, testing every component, testing every library and executable, and integrating all of the pieces together. Humans make mistakes. Bugs don't crawl into the code from the swampy surroundings, humans create them.

    While hardware logic can be tested and verified to be correct, or at least certified to meet the specified requirements and perform exactly as intended, humans and the software they create can pretty much do anything, with variations between which human is doing what things. You really can't certify ahead of time that any developer is going to produce the same output and behaviors given a defined set of inputs and requirements. Heck, you can't even assume the human generated requirements are correct in all cases. 

    The massive parallelism provided by the hardware just makes things a lot more difficult for software developers. There are specialized teams that are tasked with building specific pieces and parts of the software solution, but every team still has a great deal of complexity to tame, and bringing together all of artifacts from each team into a coherent solution is also very challenging. Parallelism at the hardware layer has been around for much longer than the majority of software developers have been able to take advantage of it. In a lot of cases the problems to be solved that would benefit from the available parallelism were few and far between.  This is no longer the case. 

    With AI the nature of programming has changed in many ways. In the past, developing software mostly involved solving problems that had a deterministic and logical solution. It was correct or it was incorrect. You could put together a logical truth table and determine for a given set of inputs all of the possible and finite outcomes. AI isn't constrained to a finite set of logically provable outcomes based on logically provable inputs. It's largely driven by probabilistic outcomes driven with massive numbers of inputs that are also subject to probabilistic behaviors, all of which are constantly evolving. 

    In my opinion, AI has moved a lot of the cognitive burden for coming up with a "correct" solution from the software development team to the software specification team that now must include mathematicians, data scientists, social scientists, computer scientists, statisticians, linguists, human factors engineers, and even psychologists. 

    It's not fair to say that Apple's software development teams, their leaders, or the leadership team are not up to snuff because they can no longer deliver software in the same manner and timeframe's that they were doing when AI/ML and Apple Intelligence were not in the picture. 

    The software development teams are a crucial part of the machine, but with Apple Intelligence they are by far not the only critical part of the machine. Software development at any level involves two high level areas of concern. The one we tend to focus on is the software team always "building things right." But the other high level area of concern is "building the right things." With Apple Intelligence a lot more of the burden has moved to the latter concern, and the "right thing" is a lot more fuzzy. Once Apple regroups and settles on “what” they want Siri to be, their software teams will be fully capable of making it happen and deciding “how” to do it. The boundary between who is responsible for the “what” and the “how” has plagued software development since day one.  
    hypoluxagatorguywatto_cobraelijahg
  • Our favorite 3-in-1 MagSafe chargers in 2025

    Nice overview of several higher-end 3-in-1 models. The only thing I would add is that some models work better than others based on where and for what purpose you are using a 3-in-1 charging station to begin with. There are subtle usability details that you only discover after using them for a little while. I've probably tried a half-dozen or more and I've found that my first impression doesn't always hold true for the long run. 

    For example, my so-far favorite charging station for my desktop workstation is very similar to the Hyper Qi2 4-in-1 in terms of form factor, but without the folding capability. The things that make it my favorite include it being rather robust, constructed with a lot of metal, and is heavy. It doesn't move around my desk or move when I place my devices on it. When I pull my iPhone 14 Pro Max off the charging stand it doesn't have any separation anxiety and releases in one-handed fashion. If you regularly use your iPhone while it's on the charging station, which I do very often, you don't want the phone to move around when you tough the screen. A robust and heavyweight charging stand does not really travel well, especially if the charger does not fold into a more compact form. Some of the things that make my stationary charging station a joy to use on my desk keep it from being a viable charging station candidate for travel.

    Looking at 3-in-1 charging stands intended for travel I'm willing to compromise on one-handed operation for the sake of lightning my travel load and saving space. However, I have found that some 3-in-1 charging stations with hinges sometimes lose the tension in their hinges and everything gets a bit sloppy over extended use. Charging stations have movable charging pads and sometimes the charging pads get loose as well. The way the folding feature works is very important. If the unfolded orientation depends of the tension of a hinge alone, rather than a hard mechanical stop, it is more likely to get loose with use unless there is a way to adjust the tension on the hinge.

    At this point in the evolution of charging stations it's an absolute must-have for me that they support iPhone standby mode. Supporting the Apple Watch's similar night stand feature is helpful, but I can overlook this feature if the iPhone can be placed on the charging station for Standby mode. This allows me to use the iPhone as a nightstand clock, music or ambiance sounds source, and wake alarm. The iPhones that support Standby mode provide settings to dim the screen and use low contrast fonts and images when the room lights are out.

    Speaking of lights, make sure that any charging station you buy for bedside use doesn't have a big honking LED pilot light or charging state light on it. These multi-device charging stations may seem like a novelty to some people, but having the ability to charge all of your personal devices with a single power cord is a great thing, especially if you're not a fan of cable clutter.

    muthuk_vanalingamfastasleep
  • Boox Note Air 3 C review: A good color e-ink reading experience, for iPad prices

    The readability of e-ink is excellent. I wish more products that currently have low contrast LCD displays would change over to e-ink. I’m talking small front panel displays on switches, routers, small weigh scales, chargers, appliances, etc., devices. I recently bought a small temperature monitor that has an e-ink display instead of LCD and the difference is amazing. 
    muthuk_vanalingamwatto_cobra
  • Apple's original and delayed HomePod finally shipped seven years ago

    I had the original full-size HomePod and I thought the sound it produced was wonderful. My original one had an audio anomaly during the warranty period and Apple replaced the whole unit after I turned in the broken one at the Apple Store and Apple shipped me the replacement. The replacement unit lasted a couple of years and it also died completely. Prior to the death of my second HomePod I received an Amazon Echo Studio as a gift so I switched over to that. In terms of my own listening experience the Echo Studio is not even close to the same sound quality as my first edition HomePod. The Echo Studio has more features, plenty of volume and bass, and a smarter voice control, but it has never matched up to what I learned to love about the HomePod. If the original HomePod sound quality was a 10 the Echo Studio is more like a 7-7.5. 

    After I reluctantly switched over to the Echo Studio is when Apple went quiet with the full-size HomePod so I've put up with the Echo Studio ever since. I've known for a while that there is a second gen full-size HomePod available, but I held off a bit because my previous HomePod and now the Echo Studio live in my kitchen and I don't spend a lot of time in there. In the meantime got a HomePod mini for a different location. It sounded pretty good for its size, but no match for its older brother. To give the HomePod mini a better chance to redeem itself I got a second one and set up a stereo pair. The stereo pair significantly improved my satisfaction with the HomePod mini. I can see where the stereo pair of minis would be a great desktop setup with a Mac, but not in my kitchen. I encountered some initial growing pains with the minis but I believe they were mostly attributed to Apple's Home app which at the time was not so great to say then least. After a series of firmware and software updates to hardware and software the HomePod mini pair has been rock solid. The HomePod mini pair still impresses me with the quality and undistorted sound, even at higher volume level. I use them as an audio output from my Apple TV, music, and integral parts of Apple Home. Those little orange buggers can really push out some great sound. I can only imagine how well a stereo pair of the big boys would sound.

    Now I'm at the point where I've decided to re-home the Echo Studio. It was free, it works okay, but it just doesn't live up to the sound quality of my deceased HomePod. I'm missing my HomePod more than ever. Now with rumors swirling about a next gen HomePod being on the horizon I'm waiting until I see what card Apple decides to play in the next round of upgrades to this product line. For my particular use case, which is room filling background music at moderate volume when I'm cooking and eating, I do not need or want a hybrid audio-visual HomePod. I need to keep my eyes focused on what I'm doing. If Apple gives the full-size HomePad a pass on the next upgrade cycle I'll gladly settle on the current version, but I'm holding out for something that is surprisingly better even if it's only a smarter Siri with a little Apple Intelligence infusion. Please Apple, just don't mess with the sound quality.

    [Edit]

    Kudos to the Apple Insider team for the recent updates to your desktop and mobile web pages. I'm really enjoying the additional controls and I like seeing the comment number badges on the top level article listings. Thank you. 
    Alex1Nstompymuthuk_vanalingamwatto_cobra
  • How the new Apple Invites app works, and when you want to use it

    This app sounds redundant and needlessly complex compared to just sending out invitations using Calendar. I’ve tried different apps and services, e.g., Slack, for coordinating and collaborating informal groups. The tools can be fabulous and seem ideal for the purpose, at least from my perspective, but a lot of non-tech-enthusiasts and regular users simply don’t want to or don’t see the need to learn yet another tool or way to do something they are already doing using tools they already know, even if they suck, e.g., Facebook. The devil you know …

    Perhaps in a corporate environment this app would be more useful, but again, there are plenty of tools and processes already out there in use for event planning. Time will tell, and who knows what other users may see in this app that I don’t see … yet. The “yet” word seems to be applicable to a lot of what Apple is springing on us now that the basics are pretty much covered.
    9secondkox2Wesley_Hilliardwatto_cobra