I think something like OAuth might help here. Modeling each "claw" as a unique Client Id could be a reasonable pattern. They could be responsible for generating and maintaining their own private keys, issuing public certificates to establish identity, etc. This kind of architecture allows for you to much more precisely control the scope and duration of agent access. The certificates themselves could be issued, trusted & revoked on an autonomous basis as needed. You'd have to build an auth server and service providers for each real-world service, but this is a one-time deal and I think big players might start doing it on their own if enough momentum picks up in the OSS community.
Vogtle won't stay the most expensive. My idiotic government (Ontario, Canada) is committing to building a new nuclear plant. $400 billion for 10GW, and that's before the inevitable delays and cost overruns. Maybe we'll break the $100,000 per kW mark!
My worst technology experience of all time was maintaining support for a Zebra label printer in VB6. I can assure you that the users of these printers had maybe 1% the cortisol response I did when something went wrong.
Designing software for a printer means being a very aggressive user of a printer. There's no way to unit test this stuff. You just have to print the damn thing and then inspect the physical artifact.
A million years ago I worked on some code which needed to interface with a DICOM radiology printer (the kind that prints on transparency film). Each time I had to test it I felt like I was burning money.
If we're worried about wasting parts of the animal, in a typical hog the blood is maybe 10% of their body weight. Significantly more waste (by mass) is created during the trimming process. If you want to do your part, you can buy a pack of "bits and pieces" whenever you need to render a lot of fat and don't really care what it looks like in the end. These trimmings are in substantially greater proportion than the finished bacon you typically buy. So, if you're making something like a big batch of beans, you can get the exact same flavor profile for maybe 10% the cost.
Serialized execution flow and large work batches seem to be just as good for humans as for machines.
Context switching is expensive in any domain once you look at it from an information theory perspective. Communication of the information almost always costs more than computation over the information. Large batches solve this.
If I'm in my kitchen and I've got everything I need to make 2 lbs of taco meat, I also have nearly everything I need to make 4 lbs. From a process perspective it's identical. The additional amount of time required is sub-linear in this situation. There's probably enough capacity for 6-7 lbs before I saturate the capabilities of my residential equipment.
I had a double take looking at the Houston map. I just got done driving that exact stretch of I45 last night and I don't know how a waymo would have safely navigated from SH288 northbound beyond I10 on I45 NB without getting stuck exiting on I10. TxDOT is considering tearing down a section of I45 in Waymo's map because it's so bad. The current negotiation protocol for a lane change on the piece elevated is threat of imminent violence during typical commute hours.
Waymo should start with the Houston burbs and work their way in from the outside. They don't have a lot of customers in this zone. Downtown Houston is pretty dead compared to most other major metros. All your customers are in pearland and the woodlands looking to visit adjacent shopping.
That specific location would probably never flood in the way that you might think. The areas you really need to worry about are downstream of the Addicks and Barker dams:
Ring doorbells and ALPRs are a meme compared to what we've already deployed domestically. I've seen the Houston police department fly wide area surveillance aircraft all day over certain parts of town. The capabilities of some of these systems are almost unbelievable.
I strongly recommend not breaking the law until you've fully considered the omniscient demigod threat model. You never know who is watching and what their capabilities might be.
This, honestly. I used to get so angry when the US subjected foreigners to treatment that no human with morals would consider, but then I noticed that, without fail, almost every thing the US did to foreigners gets done to US citizens eventually, either by foreign governments or by the US government itself. The only thing that I haven't seen yet are drone killings on US soil, but I suspect that too is only a matter of time.
Like it or not, we're all stuck on the same boat. Normalizing the abuse and mistreatment of some parts of the world means normalizing it for everyone.
reply