It's a failure if you took a 70% chance it'll flop and it gave a 200% ROI. Because if you invested in 10 companies you'd pay in 10*100 = 1000 and get back 3*300 = 900. Maybe you gambled on one company and tripled your money but then that's luck. That's essentially what venture capitalists do, they buy your business plan and sales pitch, scratch their lottery ticket and end up with a winner or a loser. Either way they're done, off to find new promising tickets to scratch. Which is why they care so much about the upside, they already know it's a gamble but they want to know the prizes. No point in a high risk gamble without a high reward to match.

Actually open source is in general massively underfunded because it has to be worth it to the people stuck with the bill - in either time or money - which is usually far fewer than those who benefit from it. Like I enjoy the fact that there's roads to drive my car on, but I've never funded a specific stretch of road. Granted, I've paid my share of taxes and toll roads but if you gave me all of that money back and told me to invest it directly into road projects I'd probably pocket the money instead. There's no place in particular I'd like to go that badly and even if I did I wouldn't get very far. And for the people with money it's more like fuck it let's rent a helicopter because it's cheaper to airlift it in than to build a road all by ourselves.

Many people have tried various forms of crowdfunding, bug bounties, subscriptions etc. but the general tendency to play chicken and see if someone else will fund it and the uncertainty involved means they pretty much all suck. As consumers we want to see the product and decide whether it's worth our money or not. But then code is already written and we got plenty other things to spend them on where you don't get to play if you don't pay so we don't actually make donations. I'm not saying capitalism is optimal either because it's definitively not but it does a pretty good job of moving money from people who want shit done to the people who can get them done.

Nobody puts people on ventilators just because they have low oxygen levels. First they put you on bed rest. Then they put you on extra oxygen. A quick search show you have people with severe COPD that have down towards 10% lung capacity but still breathe on their own. Ventilators are for people that can't even do that, there'll always be a few borderline cases but for most it's that or death by suffocation. It's a silly attempt at diversion trying to blame the ventilators rather than the "harmless" virus, when we wouldn't use a ventilator unless it's wiped out 90%+ of your ability to breathe.

How long ago do you think Apple made the strategic decision to move desktop-class computers to ARM? They'd need the chip designs, production capacity, software support, all sorts of plans for the transition. This is not a decision made in the past year, my guess is 3-4 years ago. Once they committed the resources it's unlikely Intel could do anything to change their mind.

People have been bashing Olympus and Panasonic for their "small" sensors cameras for years saying the pros only want full frame, medium format or in a pinch ASP-C. But even a MFT sensor is way bigger than the <1" sensors you find in a cell phone, there's been a market for that. Particularly with hybrid shooters who are interested in video, but Panasonic cornered that market. What I read is that Olympus tried too hard with high end bodies and glass bringing them to a price/size/weight where they'd compete with bigger sensors rather than focus on being the small/light/cheap alternative. But of course on that end you have cell phones so... maybe the path just got narrower and narrower until it dead ended.

I don't think anyone's selling as much as they'd like these days, even before corona... Canon's RF line is creating some amazing but niche gear, Nikon's Z line is okay for those already invested, Sony's innovation seems to have slowed down since 2018, Panasonic is making okay gear for the L-mount... but the most exciting stuff has actually been coming out of Fuji lately. They're not even on the podium in terms of market share but the X-T4 is impressive.

I don't think you're wrong, hell we know just a false accusation can ruin lives. That doesn't mean the police should refuse to investigate/search/detain/arrest suspects or the DA should not press charges unless it's a slam dunk conviction. While I'm sorry when innocents are caught up in that, we can't make the system such that anything we couldn't prove beyond a reasonable doubt becomes a demand for restitution. You should be able to get compensation for things that are obvious mistakes and incompetence like having a warrant but knocking down the wrong door, but not simply for being the target of a criminal investigation. I accept that a lot of guilty people have to get away to avoid putting the innocent in jail, but giving them a payday on top would be a bit much. One shouldn't lose sight of the fact that there's a whole lot of actual criminals too...

This is like saying the US has nukes and ISIS doesn't, releasing blueprints and simulation data would be a good thing. I'm sure that'd level the playing field but not in a good way. If you can't trust the cops with info, then you can't go to the cops and that's a win for all bad guys everywhere.

Just this? Probably didn't need that much, CUDA already exists for both Windows and Linux. It's made by Nvidia so they control the spec and don't need to take anyone's opinion on design or integration. All they needed to do is pass the same commands/data through to the Windows driver. Making complete drivers that actually take into account the kernel/windowing system differences between Linux and Windows will be much harder.

Yeah 32GB is just too small, I got a NUC for my dad a while back (because with his vision a big screen was a must) and I was looking for a 64GB SSD. Or actually I think was some eMMC junk but it's not spinning rust. It had only 2GB of RAM but I could pop in an extra 8GB SODIMM which meant that despite the channel mismatch it had 10GB of RAM total so you wouldn't hit the storage very often. The user experience was actually considerably better than I expected. If the RAM had been fixed at 2GB.... hard no.

Uh, you do realize that if they'd successfully amended their bid that means one of their competitors would have been left with nothing right? This is not a victimless crime. It's massively tilting the board in favor of one company, unless you'd also tell the others how to beat Boeing's proposal.

Or in an alternative reading, you make a sham competiton then pick the company you favor to win. There are processes to amend/restart procurements giving everyone equal standing, sidestepping the whole process to award the contract to your buddy is not one of them.

If he'd actually succeeded in feeding Boeing inside information and the contract had been awarded to the wrong party that'd probably be prison time and many millions of dollars in liability. His goals might be noble but his judgement is highly questionable. What else is he willing to do to keep/accelerate the schedule?

Except for the money. Apple is already investing heavily in CPU/GPU development for the iPhone/iPad. If they could use that to create an in-house chip instead of buying from Intel - who is not known for their low margins - they could pocket all that profit themselves. I agree that consumers don't want ARM or any CPU archtecture in particular, what they care about is applications so the key is vendors. WinRT had no commitment and no clear future, it was a niche product by Microsoft to plug a percieved gap in their lineup but there was no indication they'd migrate away from x86 meaning the business case for making an RT version of your software was terrible.

If Apple does this it'll be with the commitment to migrate at least their whole consumer line of Macs. If you are producing Mac software you can either be there or... not be around much longer. That's a pretty easy boardroom decision. That said, I expect it'll be everything /.ers would hate like locked bootloader, store only apps. But Apple got the financial muscle to make it into a compelling offer for most people anyway. Cue the people saying it'll be totally useless for them, I'm sure that's true I'm just not sure it'll be in numbers that Apple cares about. They've found that professionals run a pretty tight ship of cost/benefit if they want to stay in business while consumers pay based on disposable income. The most profit is in selling bling to people who can afford bling.

Even a toddler has had years worth of dual high resolution video feeds, experiencing motion, ego-motion and interaction with objects in a 3D space. For an AI those pictures are literally all it knows about existence, the equivalent of a toddler that's been completely paralyzed since birth, blind on one eye and sedated while cared for so the only visual stimuli it's experienced ever is that slideshow. It's not an apples to apples comparison.

We've recently been making great strides in unsupervised pre-training. Basically we let the model study a ton of unlabeled data first, then it turns out we need only a few labeled examples/counterexamples to identify a class. The reason you need counters is that you don't know if the class you're trying to find is "insect" or "bee" so we also need some information on what's not bees.

Of course that's still only a fraction of the human mind, but it's hard sometimes to distinguish between reasoning ability and information compression. The more you're able to reason about the world, the more you can derive, the more compact your representation of the world is. If that was a scientist turning raw observations into a formula we'd call that intelligence, if it's a computer creating a compact representation through decomposing/deconstructing a complex result it's not.

Even if that was true, I wouldn't announce it at the same event. Apple would want all the buzz to be about their new ARM products and portray x86 as becoming a legacy system. Then later spin it as falling out of favor with Intel, but don't worry with AMD we still have our existing customers with x86 needs covered. Because I have the feeling when their ARM line comes it'll have "courage".

I think you underestimate the number of odd needs around the world. Like if all the world's oil platforms and nuclear plants bought one each for inspections they'd have their 1000 customers already. They're not pretending it's a mass market product, the user manual is pretty... direct:

Top of page 12 if anyone wants to double check.

Well 90 mins runtime (no payload) * 1.6 m/s top speed = 8640m = 5.37 miles. Add an extra 4.2 kilo battery as payload and you should in theory be close to 10 miles (one way) and still have 10 kg usable payload. But I'm guessing that's doing laps on a flat race track, if you're actually using it in nature my guess is you'll get nowhere near top speed or that power efficiency. Plus you have to deduct for payload, they don't give numbers for that. Most likely it can carry a candy bar for you.