Sadly, even self-hosted software isn't immune from the exact same problems as connected apps. You have a degree of control with something self-hosted, but once it's got a couple of CVEs against it, you're going to need to upgrade - and then you get the bloat and enshitification.

I don't know what the answer is - keeping away from Silicon Valley is probably a good idea, and open source is an option I suppose. Buying from companies that aren't going to IPO any time in the next 5 years or more, perhaps?

FWIW, I have 'smart' heating control, made by a small company in the Midlands (of England). It's what it is - it doesn't get glitsy new features, it doesn't even get many bug fixes really, but it keeps on chugging away as well as it ever did. It still needs hardware upgrades from time to time though, so there's an element of "subscription" to it even then. I was an early adopter, and man, I suffered for my choices in the early days - but these days, it works pretty well.

> Why would any rational person choose this?

Because they were promised "better". They were told that, yes, sure, the vendor has inserted themselves into lots of things, but that vendor was super keen to make sure it all worked out really well and that things would be great as a result. They were going to make everything talk to everything, and it was all going to be easily controlled and managed.

You might fall for all this as a rational person. As a cynic, you never would. Some people aren't that cynical.

FWIW, home automation is cool - but as far as I can see, no one vendor does anything like a decent job of it. If you can't have multi-vendor, multi-platform, mutli-everything, then it's going to be a bit limited/crap/compromised. If you can have multi-everything, then it falls to you to do all the integrations and make it all work - so nothing like the "plug and play" utopia the vendors tell you they can deliver. Even then, swapping one vendor for another isn't always easy, and actually a lot of supposedly "smart" devices are really pretty much dumb as a post.

Whatever it is about home automation, it seems to me that no one has "cracked" it yet - nor does anyone seem to be on a path to do so. If anything, most vendors have realised it's not the money making cash-cow they thought it was and are scaling back their former ambitions.

Time to get a heat pump. At least that way we're sucking some of the heat out of the atmosphere to warm our homes rather than burning gas to make even more heat.

Yes, the electricity might not be very green - true, but that's a solvable problem, and is getting solved.

It opens every door - AND it opens the deadbolt too. The only thing it can't open is any 'chain' on the door (because that's nothing to do with the lock - and so not always available).

One thing about housekeeping keys - they don't typically open the deadbolt. The manager's key likely does, but there are a lot less managers than housekeepers. Suffice to say, this is a pretty catastrophic flaw, and reading between the lines requires some new hardware to resolve in at least some cases.

> Redis' parent company just signed its own death warrant.

They may well have done, and that'll be a shame - they've been good custodians of it until this.

A forked version (can we call it Radish?) seems likely. Even Redis Inc have struggled to invent any new features, so the fork will likely be plenty for most people. The plugin system in Redis 7 might see some clever stuff getting integrated, but the likes of the cloud providers would most likely never support such a thing anyway, so they can carry on serving up "vanilla" Raddish.

As a side note, I can't imagine ever needing any paid support for Redis - people obviously do pay for it, but I wonder what they get from doing so?

Pricing is hard - if you put it too low, you don't make as many billions. If you put it too high, you also don't make as many billions, and you look like an utter dick.

Looks like they went with the second option.

When I last looked at this, the traditional ELK stack wasn't so easy - they're trying to take a lot of stuff "back in house" and are making licensing difficult as a result.

I'd personally suggest Graylog (even if they can't spell "grey"). It's not really better or worse, but it seemed like it was a bit closer to the free world than ELK. There's another (very much open source) option, which largely runs in a single process - the name escapes me, but it's not really ready for prime time yet, but it's definitely interesting.

As for Splunk, they worked at massive scale (if you wanted it), had all sorts of nifty enterprise friendly features and they were the market leader. I'm not sure I really understand the Cisco fit very well, other than "to get the client list". It was good, it was probably beyond the reach of anyone smaller than "pretty big", but I'll miss it when the inevitable enshitification starts.

Every Tesla taxi driver I've asked has said they've had to get fixed several times, and their car isn't 3 years old yet. It seems high utilisation kills off Teslas pretty quickly - not something perhaps seen by normal-mileage Tesla owners who maybe get a couple of parts fixed during the annual service and think that's perfectly fine.

Accidents are probably much higher in rentals, but if the rental is actually doing what it should and getting high utilisation, it seems Tesla's terrible build quality will have an effect too. We'e heard plenty on slashdot about how Teslas tend to get written off for relatively minor accidents, so I'm sure that adds to the costs too. Hertz bought a lot of Teslas because that's the car people want to rent - the crappy Leafs probably don't have as many problems, but no one wants to rent them.

My take is Hertz bet big on *Tesla* (rather than EVs in general) and that didn't work out so well for them. That's not necessarily a reason to avoid Teslas for self ownership, it's perhaps more of a statement about Teslas limitations though.

My prediction... AI will never work for shopping.

I reason this because no store will ever make an AI the users will ever want to use. Stores will always try and push what works for them - not what works for the customer. Taking a simple example, lets say I need to buy some AA batteries. The vendor will always try and push me to whatever makes them the most money or empties their warehouse the quickest, so they'll sell me a big box of batteries, or they'll sell me premium brands, or they'll sell their own brand and try to hype them as being as good as the premium ones - whatever works best for them. Whereas, I would like to buy a smallish pack (so I don't have to spend so much money, and so they don't "go off"), I'm not really bothered by the brand, so long as they actually work well and keep their charge for a long time. Once I've figured that out, I'm price sensitive, so I'll buy the cheapest of the "good looking" ones - whatever any of that means, I'll compare deliver times/costs, etc. Either way, it's almost certainly not in the vendors best interest.

Could a third party (eg. a comparison website) make an AI that works for customers? I suspect probably not for all the same reasons as above. Ultimately, the aggregator relies on the vendor, and the vendor will be pushing their own agenda, so the aggregator will be tainted by that too. The aggregator needs to make money, and so they'll introduce (or persist) biases just like the vendors do.

Could an open source option surface? Maybe, but it'll be beholden to vendor data, so would probably struggle like the aggregators would.

Aside from the obvious vendor biases, I suspect the buying decision is also actually insanely complex - I'm wondering if it might be a problem like self-driving cars - it looks simple on the outside, but it proves to be really, really difficult to get right. If you've ever seen someone pick an apple at the supermarket, you'll know that some people seem to have a lot of unpublished, and largely irrational variables in their buying decisions. Getting all of that right seems like it'll be very, very hard.

All true - but if they were to commercialise and actually make it good, then having a cluster of machines that wake up at the right times, run jobs and report results could be a handy feature of many application stacks. Of course, we all do the same already, except our current solutions all suffer from problems if things crash, or if inputs are out of range or whatever. We've glossed over and 'normalised' those problems for so long we probably don't even know we have a problem that could do with a decent solution.

There's obviously a lot of 'hypetrain' going on here, and there's no specific need for a whole new OS to solve the problems they're solving, but if they make a decent solution, then it's still a decent solution. I rather hope they don't over-cook this whole thing and actually come through with something, but the cynic in me thinks they'll do the usual VC-backed hype-to-the-max and then produce nothing of any real use.

https://developers.slashdot.or...

As last time... it's good, probably better than any other LLM at writing code and doing technical tasks, but it sure as shit isn't as good as you're making out.

This lot have some bold claims - namely that it can produce entire programs, rather than just a few functions (their example is that it can write simple games). Also, they claim that it's done some of the jobs on Upwork.

Personally, this one needs a good deal of salt. The claim about Upwork is likely only true after you've spent some time talking to the client to work out what they meant in the job spec and then maybe iterated a handful of solutions until you got something that looks about right. The thing they likely have right is the quality of the output will be similar to a lot of Upwork responders ;-)

To boil this down, I think they've made a super-focussed model that's good at coding tasks (and probably terrible at telling you about celebrity gossip). They've presumably coupled that with some ability to test or evaluate the code it generates to see if it meets the brief - that's quite cool, if it's anything like I've described it.

MH370 is something of a modern mystery. It's also been a significant 'wake up' call to remind us that we don't know everything - even in something as heavily monitored and tracked as an airliner. That it seems to have disappeared so completely, and even after this much time hasn't given up its secrets is really the stuff of conspiracy novels.

I really do hope it is found - but my guess is it'll be found by accident now, rather than because anyone was specifically looking for it. By then,

All probably true, but this is something akin to clearing out the typing pool. That is, a big business has a handful of people who "do cashflow". They work out how much money needs to be in any one bank account at any given moment. Any spare money gets put to work in some way or other - some of that work is long term, some is short term. The cashflow folks job is to make sure there's as much money being made as possible, but that they can always cover their payments when they're due - so it's a juggling act of moving money around different accounts, moving it into investments of different types and then pulling it out of those investments when they mature, or when the money's needed somewhere (accepting there'll be a financial penalty for pulling out of an investment early).

This is absolutely an area where "AI" can help - most likely some ML and statistical modelling (rather than an LLM). In fairness, JPMorgan are in the process of productising something that's very specialised, quite hard to do (because it's specialised, and because there's no "dev environment" to try it on), and potentially could make businesses a bit of extra money, and save a bit of headcount.

How that means entire businesses can go to 3.5 day weeks I'm not really sure. My guess is humans leave a lot of money in accounts that doesn't need to be there (humans thinking "just in case" and "CYA"), where computers have much less worries on that score. As such, the computers can move more money into investments, and can probably predict the time the money will be required more accurately (ie. again without the fear-factor overhead a human would add). All that means more money for the business, but I'm still not sure how thousands of people would suddenly have more leisure time as a result.

> What users look for in a battery

Well, we have regulators, who represent users. And they rather like "recyclable", because they look for the whole product to be recyclable in some sort of way because we don't like nasty things in our landfills, and we recognise that we can't just ship it all off to the developing world for them to have it in their (unmanaged) landfills instead.