Too expensive
They're nice little gadgets when you have a use for them, but the prices are always steep for what you actually get.
When you need that specific niche filled great but otherwise it's difficult to justify the cost.
355 publicly visible posts • joined 16 Jun 2007
That 'mini dstacentre' would be what, half a racks worth? Maybe a whole rack if you really wanted to push it and have lots of spares. Hardly extreme.
And you don't treat the possibility of no aircon by having no aircon; you have aircon, you have backup aircon, then you have contingency cooling. You don't just skip to the contingency straight away.
Not a big shock though, often the team budget for these things is actually derisory. And no one actually wants the jobs as the pay and conditions are rubbish as the expectation is that 'working in F1' is a reward in itself. Hint; after the first week it isn't.
They might shift more product and make more money overall if their prices relaxed back a bit, currently they range from steep to utterly insane when you look at the proper good stuff.
And that's before you get to the double-dipping for things like vGPU. That leaves a really nasty taste, charging to use hardware you already paid a fortune for.
And when you consider the mostly minimal improvements in performance with time and what seems like a slowdown in product iteration it feels like they've got lazy.
Shame, as my need for fast GPU is growing by the day for all sorts of things.
Seems like a complex expensive resource intensive solution to a problem that doesn't really exist.
I do wonder what they expect to feed their bioreactors on 'cos it won't be grass. And what happens to all the land used for arable farming?
One thing they may need to think about is where they're going to find cows to get their ongoing biopsies from if this goes mainstream; no one is going to keep cows for fun so commercial herds and breeds would go extinct, just like the specialist breeds of agricultural horse did once machines took over, or many fruit varieties. Commercial plant and animal breeds can disappear amazingly quickly.
Interesting tech but no panacea. Still probably healthier than that pea based muck with all the saturated fat thrown into it.
Another aircraft program that shall remain nameless has a Single Harmonised Integrated Test Environment for all development work. The title is displayed in an extra large font with emphasis on the first characters just to make the point clear.
The Brits put in quire a few things like this (plus various scattered insults/digs), not sure the partner nations always get the joke.
When you start looking into how neuromechanical control works you realise there's a lot more to it than the brain, muscles and some wiring in between.
A lot of muscular control and sensory feedback handling is embedded in central pattern generators in the spinal cord. Lots of autonomous actions and interactions that don't even need input from up top. Stuff like leg and foot motion, motor response and gait is driven by this layer not the brain.
So hopefully they've put their work into detecting intent and built the fine control and feedback into the hardware itself, because that's how the original systems work.
Rotating out to some other job is a pretty normal thing for his career, and at least he gets to do some flying (other than a desk), but I can't think of many fast jet pilots who'd see playing with an antique 747 as a great pastime even if they might get to shoot fireworks off it occasionally.
If you really want proper accurate hand and finger tracking you need quite a few IMUs (16 or 17) to capture the full possible range of motion. You can get away with smaller numbers if you're prepared to compromise on what you can capture (ie much more than fingers flex).
Of course this only truly matters if you're doing something that needs that level of fidelity.
The other issue is usually calibration, especially if you're trying to correlate with touching something physical. Then you need to measure that hands at some point to get the skeleton right and make sure you've got a decent hand waving calibration phase. With haptics instead you probably don't care quite so much.
These things seem quite expensive for what they are compared to other gloves with similar motion capture so what you're really paying for is the haptics. For the money I hope they've got the durability finally sorted!
These things are great when they work but we're still at the stage where there are lots of entrants in the market and none of them are perfect yet. Hopefully there's enough market demand to get this to settle out.
Leaking the data is sloppy.
But the other question would be why they have retained so much for so long? Surely after the parking is validated and after a suitable delay for any challenges (like fines) they should be binning it? If they want long term statistics they can process and anonymise it and not need any of the source data.
Certainly no need to store all those images and related data permanently and risk them leaking.
Reports on Twitter (eg. https://twitter.com/Astro_Jonny/status/1168592399729397767 ) that ESA contacted SpaceX to ask them to move their satellite and they said no and were generally unhelpful.
Whether this was because of their Musky corporate culture, or because their cheaply built junk satellite is actually incapable of manoeuvring isn't clear.
That's going to be a good trick - the bulk of the cost difference is in the battery and given those are already (mostly) a commodity item in mass production where exactly do they see the reductions coming from?
You'd have to see all sorts of innovations appear and material prices collapse for anything significant to happen. And even then it still won't be cheap for a 30-100kWh battery.
The sad truth is that people will soon forget all about this, just like they've forgotten all the rudder related issues of the 737.
And just like the big aircraft suppliers have forgotten the past and are firmly back into the 'do it cheap, do everything in house, everyone else are idiots' ways of thinking. Over the last 10 or 15 years everything has become extremely parochial again.
So they're basically saying that their current working but superceded product is going to be frozen with the current feature set? But they'll still provide security updates? And they have a replacement version with a better feature set?
That isn't abandoned or ophaned, that's a normal product life cycle. Abandoned is when it gets no updates at all and the whole concept and infrastructure is binned and this isn't that.
What exactly were people expecting to happen with v1.0 going forward?
So basically you could do something because there's a non-encrypted bus, but it's a complicated way to achieve something you could do much more easily in other ways with the same level of access.
It's a bit of a common theme with security research to state the obvious and look for problems that really aren't.
So lots of warnings about all the critical devices that might be affected, but the actual VxWorks versions intended for those things aren't affected?
So are all the dire warnings true or not? They seem to talk worst case a lot (scope of effect on important kit and difficulty of fixing certified kit) while also letting slip that this sort of thing should never have been impacted anyway.
A certified RTOS is a special beast for all sorts of reasons so I'd really hope no one used a normal version in anything that mattered.
More because if they really let it out in the wild it would be even more obvious that it spouts even more gibberish than a Buzzfeed reporter - it's only convincing if you don't read very far and are really bad at spotting the inconsistencies.
There are better systems out there.
At what point do they discover what's actually hiding behind the usual Musk smoke and mirrors?
Still waiting to be convinced they have anything new, interesting or innovative compared to the rest of the market. At the moment they just seem to have a spin on the usual old stuff and even that feels thin once you scrape off the hype; GPT-2 certainly didn't seem like anything special or even particularly capable.
Having created conduction cooled boards in the dim and distant past they tended to be a bit thicker than that to get the copper and heat conduction layers in. Plus they were usually firmly strapped to something at the edges to dump the heat into, pcb foil layers not exactly being renowned for having massive heat capacity. So I'd assume the 'heatspreading' on this isn't adding much at all.
Hope something useful and cheap pops up to sort this out.
All involved in this are idiots, Hopkins for being Hopkins and the rest for wasting energy on reacting to her. Just ignore it and let it die, it lives on the reactions it gets.
Impressive number of words dragged out in response to nothing though so bravo on meeting quota.
They're complaining that the extra money they get instead of a company car allowance they're no longer entitled to is somehow a problem because in certain circumstances it goes away with the rest of their salary?
Just think, they could have just lost the car allowance and got nothing at all.
I'm also struggling to understand exactly what the proposed solution would be?
You'd almost think software testing didn't exist to pick up the bits the coder missed.
Then again the modern paradigm doesn't really consider proper testing to be a thing. And QA, what's that?
Safety languages are like safety scissors; a blunt tool handed to those incapable of dealing with the risky version, at the expense of much reduced ability and performance.
And it's not like the risky areas of ' unsafe' languages aren't well understood, it's always the same handful of basic errors.
Let's take a wild guess that they started off with a shopping list of vague requirements and a naive understanding of what was possible.
What's certain is that anyone with a brain wouldn't try to build an emergency sevice comms system on top of a 4G (or any other) phone network. You build it with technology that is actually suited to the problem.
If you want fancy smartphone features *and* a decent rugged comms system then you do what everyone already does; buy two handsets and take advantage of normal cheap phones for that side of things.
The priority now should be to rebid rather than throwing more money into a bottomless hole.
Am I alone in being slightly underwhelmed by their efforts?
They strapped a missile launcher to the engine ferry point so not exactly a huge amount of work to the aircraft, and launching missiles from under aircraft wings has been a thing for a good few decades now.
Even the X15 was launched the same way!
It's interesting and all that but not exactly pushing the boundaries of space science is it?
You'd almost think that people keen on selling or renting out big piles of kit had an interest in promoting a vision of 'AI' that involved brute force via big piles of kit.
Shame really, some of the most interesting stuff including things that people would considered closer to truly 'intelligent' systems doesn't rely on big piles of hardware. It's not that a crude version couldn't demand an impractical pile of kit but that's where clever design comes in and gets you stuff that runs happily on a Pi. Check out what a genetic fuzzy tree can manage for example.
Taxing on revenue rather than profit is a stupid regressive idea for all sorts of reasons and should be stomped on before it spreads to everyone else.
The better solution would be to simplify the tax system and either block or render unnecessary all the miriad ways of hiding or moving profits that are open to those who can afford the teams of lawyers and accountants.
But why go to the effort of fixing the real cause when something simplistic will give you a warm feeling so much more easily?
Once upon a time they would have treated the nominal cost of free or discount licenses as a worthwhile way to ensure their products were widely used and as many people as possible were familiar with them, ensuring they won the larger market.
Now they screw the pennies but ignore the long term cost.
Not a huge surprise considering they also 'save' money by cutting back on QA (which once used to be actually not too bad) while forgetting that selling an unreliable product ultimately costs you sales.
No need to name names as it's pretty obvious who's responsible!
Alexa definitely triggers without needing the keyword, it does it all the time and doesn't even reliably record that it responded.
It's pretty reliable at piping up in the background during phone conversations but my favourite so far is getting it to trigger in response to a particularly loud fart. It was confused by the question apparently.
This lazy bit of work makes it look like there was a simple leak in the hull and fnar fnar isn't that funny and look how stupid the builders are.
Other people might have mentioned a burst high-pressure seawater supply pipe, damage to bulkheads and deck plates and 250 tons of water in a compartment or two. Potentially serious, and slightly more complicated.
Detail is the difference between proper reporting and lazy hack scribbling.
He also claims that all cars have been built with the new hardware for the past couple of months. Yet customers are still getting new (June built) cars with the old hardware.
For any Musk statement assume a 30% chance that it contains any truth at all, and then assume that if it contains any truth it will be something other than you expect.