Walking into a meeting at Kohler and keeping a straight face after saying the words “toilet camera”–and then maintaining that straight face throughout the product’s discussion, greenlighting, development, manufacture, and sale–deserves an Oscar.

Oh, yeah. I think if it was a problem I was willing to spend any more money on than I already had, I could’ve potentially ended up there.

The porch lights in question are actually string lights, and I just assumed that the power outlet they’re plugged into was too deep on the porch for a light sensor to be reliable. I could definitely be wrong, though.

(Side note, I just realized I said the times slightly wrong. We actually wanted it on at 6:30 and off 15 minutes before sunrise.)

Yeah, I had it turning off before sunrise just fine. The problem is that we didn’t want to turn them on until 6:30, but on the longest day of the year, sunrise actually happens at 6:14, which means that the lights would get the signal to turn off before they got the signal to turn on, which would mean that the lights would stay on all day until the night automation turned them back off again at 10pm. Which…probably doesn’t make a difference, but it would bug my totally-not-neurodivergent brain.

Anyway, I don’t use Home Assistant, but that’s probably the one I’ll choose the next time I move.

Or, wait, are you saying that my original comment that you’re replying to is ChatGPT? Because…lol, sadly, no, I’m just like this. “This” meaning pretty much everything I write is way too overwrought.

My first draft of this did mention that there was a version of the second type of IT guy who cobbled everything together with workplace castoffs and conference swag, but I couldn’t figure out how to make it work without just being over-wordy.

I spent way more time than I care to think about figuring out how to get my porch lights to come on at 7am and turn off 10 minutes before sunrise without breaking when sunrise happened before 7am. I tried some serious Rube Goldberg nonsense in multiple iterations, until finally I decided to just add another “turn off the lights” at 9am every day. Most of the time it doesn’t do anything because the lights are already off, but on DST day it accomplishes my goal of making sure they don’t run all day, since 9am is always after sunrise.

I figured it was made up (“@it_unprofession” probably ran out of content ages ago), but it doesn’t look like actual AI content to me. The sentences are too short, for one thing.

Looks like it’s MIT-licensed, so it’s probably time to make a non-Anthropic fork.

I’ve only ever met two types of IT professional. Either:

• Their home network is immaculate and smooth as butter. It connects quickly and integrates with everything. They can manage it all from their phone, but they don’t have to because it’s all automated. Their server room (a) exists and (b) is cable managed. There’s a wireless access point and connected smart speaker in every room, including the garage and the back patio, but they’re carefully located for maximum sound coverage and to prevent signal interference. Their home theater is substantially better than a movie theater, and their media server is packed to the gills with content. Network security is hardened, with bespoke subnets for every user and tunneling for the media server and smart home functions. You feel a sense of calm and ease when connected to their network. “Everything I do at work, I try out at home first.”

Or:

• Their “home network” is a single Belkin router from 2011. They’ve had it since college, and it takes 9 minutes to reboot (which they have to do daily). It doesn’t even have Tomato on it and still uses the default password. They still watch OTA TV and Blu-Rays, so the wifi is exclusively connected to the smart switch that their tea kettle is plugged into so they can start their hot water before they come downstairs. You feel guilty even asking for the wifi password. “Why would I do any network stuff here? I do IT all day at work, the last thing I want to do is even touch a Cat5 cable at home.”

I mean, counterpoint: paid, enterprise editions of Linux have existed for decades without destroying the libre versions, and as long as it’s fork-able, it always will. But increasing adoption, even of the “captured” versions, will benefit libre code lines; in obvious ways, such as when enterprise developers commit their changes upstream (Google still does for Android and ChromeOS); but also in less obvious ways, like making development for Linux more attractive (resulting in more compatible software) and in making the OS paradigms more familiar to the layperson and the junior dev.

It was for me about ten years ago, but because of that brief dalliance, when Microsoft really finally started running toward this particular cliff last year, I was already familiar enough with Linux to be comfortable diving in completely; I don’t have a single Windows install in my house anymore. So it’s not always for nothing.

Even if the Windows voice experience put Jarvis to shame, I wouldn’t be interested. I don’t want to use voice control on my computer. Just about the only time I actually need voice control are when I’m far away or my hands are busy; so it’s nice for turning lights on and off when I have my hands full, or controlling timers when I’m cooking, or turning music on without getting up from the couch. Sometimes I’ll use voice-to-text if I have a lot to say or need to think it through. But I almost never want voice control (even if it were completely perfect, which it is not!) for the same reason that I listen to podcasts on earbuds: I don’t want to bother other people! Certainly not while I’m working, and definitely not when it’s liable to take agentic actions for me.

Buttons, knobs, levers, sliders, keys—all of those are better than voice control 999 times out of 1000. I don’t even like touch screens that much, and I’d prefer them over voice control.

The Microsoft executives inhabit a different reality than I do.

Sam Altman told reporters at a private dinner in August that investors are “overexcited” about AI models and that “someone” will lose a “phenomenal amount of money.”

That feels almost like taunting.

Depends on when you load it. They refresh every minute. In the first one I got, Deepseek’s was almost functional, but Haiku had one that was surprisingly good.

I’ve never seen any of the OpenAI models come up with anything that was more than completely broken.

This wouldn’t be a problem if they left the seats where they were from the factory instead of squashing them all as close together as possible.

The suit doesn’t say that they paid extra for a window seat, it says they paid extra to pick their seat (that is, to not have it randomly assigned). So they paid extra to select their seat, but the selection they made did not include all of the relevant information.

No, they’re saying that some hardware manufacturers report 80% as 100% (as you noted) while others do not. Just like some manufacturers report 5% as 5% while others report 10% as 5% with the realization that most people misjudge when they’ll be able to charge.

Well, the market will definitely contract. I would say at least one of the big AI players will go out of business or be acquired by a competitor over the next few years, and at least one of the big tech corps will sunset their AI model over that timescale as well. Nvidia stock is going to take a steep nosedive. I think the future for consumer AI is mostly in small, quick models; except for in research and data analysis, where just a few big players will be able to provide the services that most uses require.

They currently have enough money to keep going for a while if they play their cards right, but once investors realize that the endgame doesn’t have much to offer them, the money will stop flowing.

I’m probably going to be allowing most of my streaming subscriptions to lapse over the next year or two. Gonna stick with Dropout and PBS, but that might be all.