Does it matter how long it takes? Doesn't just total output matter?
Heck, "throw more servers at it" has been an accepted practice. Who cares if it takes 30 hours as opposed to 10 if you can have 3x the number of cars under construction at one time.
First of all worker 30 hours of labor * factory workforce is a significant fraction of the cost of a car, and not easy to improve.
Additionally a plant that takes 30 hours per car is typically going to have less output than a plant that takes 10 hours per car. Typically each minute during production the car is taking up space, being worked on humans or robots, etc. Additionally efficiencies favor shorter pipelines. Generally a 10 hour pipeline is physically shorter than a 30 hour. Parts, staff, troubleshooting etc is easier on a smaller pipeline. Additionally having each car bridge 4 working shifts for the workers has overheads of it's own compared to 1-2 shifts. Doubly so if you are trying to iterate quickly.
Imagine an assembly line moving at 0.5 mph, for the same physical foot print you could have 3 lines that are 10 hours each instead of a single line that runs for 30 hours. Now imagine there's a issue that shuts down a line for an hour, it's better if it's 1/3rd of your capacity down instead of 100%. If you try an improvement, it's better to put 1/3rd of your production capacity at risk instead of 100%.
For a typical car, only about 10-15% of the manufacturing cost is labor. And that's talking about Toyota Corollas and euro econobox hatchbacks. For a premium electric car, it's probably a single digit
Also, the whole claim would really need a citation, and careful parsing of the methodology. 3x difference in a mature industry like car assembly is very unlikely, and requires a very strong proof to back that statement up.
Tesla gives tours, and frequently discusses optimizations in reduced part counts, design tweaks for manufacturing, etc. I'd think someone with some expertise could tell if Tesla's claims are true.
Some highlights: front 3rd of chassis is one piece, rear 3rd of chassis is one piece, the glass roof is a huge piece and before that's installed things can be installed by robot (steering wheel, dash, seats, etc). Many things are radically simpler, like the interior, dash is basically a aluminum bar the width of the car with an arm for a 15" monitor. Even the ventilation system is simple, few parts, and easy to install. Most cars spend quite a bit of complexity for gauges, displays, buttons, vent controls, temp control, fan control, fans, emergency brake, air valves, etc.
Monroe on youtube (engineer with significant experience, used to work at Ford) took apart an early model 3 and wasn't impressed, didn't recommend it, and called out many details that were expensive, labor intensive, etc. He did a new model 3 and model Y and left quite impressed. Many innovative changes to make things easier to produce, cheap to reduce, reduced parts counts, etc. Monroe claims tesla is iterating their production faster than anyone else he's looked at, and he's looked at a very wide variety of cars (and other products for that matter).
Monroe also reviewed the VW ID.4 and left quite under impressed. Ironically he didn't expect much from the Ford Mach E, but was very pleasantly surprised to see innovative engineering, impressive design, and overall was quite impressed. Although he was horrified by the complexity of the cooling system and has serious doubts about it's long term reliability.
30 hours does not necessarily mean 30 man hours. And part of the difference seems to be that parts are assembled before they get to the line with Teslas (the front and rear 1/3's) , so it could be that they have the same labor costs baked in there.
Meanwhile, I assume that space is the cheapest part of the car factory, dwarfed by robotics.
I mean, the VW is cheaper than the 3, so that implies they aren't spending that much more.
I’d be curious to know what’s QA in that breakdown, or where you got that data, and what Tesla’s numbers have been over time.
But let’s assume that’s true. Tesla has been building electric cars for how long? VW is basically just getting started for real (modular platform for all their cars unlike the eGolf), and still makes ICE cars as well.
It’s never a good idea to discount your competitors. I doubt Elon does even if comments on HN do.
The economics of scale trickle down into cost savings for the consumer, even if it is by only a fraction of what the manufacturer saves. There's a reason the id.4 costs about the same as a base Model 3 and you get less of a car for it (not factoring in federal tax credit, which might be returning to Tesla soon anyways).
The manufacturers no doubt. Musk, and by extension Tesla, has been seriously focused on ease of manufacturing. Probably since the huge problems Tesla faced in the late 2010s.
They should be. The longer the build cycle the more errors there will be in the build.
Each job is a dependency for the next job. A long build time either indicates that individual assembly technicians have too many steps in their job (it is much harder to remember 25 steps than 5 steps) or that the plant is assembling more parts locally than their competitors. If you have an engine shipped in complete from your supplier that is less time your plant spends assembling a dependency at build time.
Electric cars should in theory assemble much faster than gas. Fewer parts. Less complexity in the mechanical assembly of the drive train.
When corporate overlords dictate that an assembly plant meet production numbers that it isn’t capable of accurately producing they shell out complete garbage. Inspection lines flag errors that managers ship with disregard in order to meet numbers. Every time.
Teslas are much simpler designs to build at scale, while traditional automakers have perfected the ICE design and manufacturing, they see everything through that eye and bring the warts and complexity from the devil they know (as an example, most non-Tesla EVs do not have a floor battery architecture). It is not at all inconceivable for Tesla to leapfrog the traditional automaker in building EVs they design. That does not contradict VW's ability to produce ICE at scale.
Interesting, I haven't tried Jami. If I am understanding it correctly it seems to provide calling features, which Zoom and other typical video conferencing type applications don't.
I'll also through Jitsi out there as a very capable FLOSS alternative to Zoom. If you tried it a few years ago it's changed dramatically in the last while. It's now WebRTC based and runs in browser without any download. It's not quite end to end encryption because the stream needs to be decoded on the server before being re-encoded for the other clients, but since you can easily self host it on a cheap VM I find this acceptable.
Jitsi does not re-encode video streams on the server, it's a SFU (selective forwarding unit) and just forwards (some) video streams to the other participators.
When Simulcast is enabled, multiple resolutions of your video are streamed to the Jitsi server and it only sends those streams to clients which they asked for (participants can choose the quality of video they receive, in the UI)
Can’t WebRTC support peer-to-peer video calls without running the stream through a server? I have vague recollection that it can but am not particularly familiar with the protocol.
To answer my own question: Yes, WebRTC supports peer-to-peer video calls but does require a “signaling server” to help establish and close the peer-to-peer connection. [1]
The signaling server is only useful for the connection initialization though, and it never has access to the video stream, only to metadata (like your IP address, the supported encodings of each party, etc.). And it doesn't even really need to access them: it just needs to forward them from one peer to another, so it could be end to end encrypted.
It also needs one or several STUN servers as part of the hole punching scheme, but this one doesn't even exchange anything with anyone, so there aren't many issues here (and you don't need to roll your own: you can use Google's one)
TURN doesn't have anything to do with number of participants; it's there for when NAT hole-punching completely fails and you need a relay. TURN is application-protocol agnostic and just forwards packets; it does not need to decrypt whatever it's relaying. Now, one could run a malicious TURN server that MitMs connections, but I'm not sure how obvious that would be to the end-parties.
For more than two users, you mainly have three options:
* MCU (Multipoint Control Unit), which IIRC does need to decrypt your video, as it will post-process it and possibly re-encode it to send a single stream to the other participants.
* SFU (Selective Forwarding Unit), which in theory doesn't need to decrypt your video, but does need some metadata about it in order to make smart decisions as to what streams to forward to whom (for example, forwarding only the stream of the person who is talking). In practice, I believe some (many?) SFUs will do decryption, thought it's not a strict requirement.
* Dumb peer-to-peer-to-peer-to-... multi-forwarding. You can of course theoretically stream your video to each other participant, and they can all do the same, but that quickly fails to scale. It might be ok for three, maybe four participants, but even then there will likely be problems.
1 - Kids' desktop (Raspi3)
2 - Kodi (Raspi2)
3 - LAN print/automated backup server (Raspi1)
I've order a 4GB Raspi4 to upgrade the kids' desktop and 1 GB Raspi4 to upgrade the automated backup server (Gb ethernet + USB3!). Kodi will get the old kids' desktop.
Another data point, I have an XPS13 L322X that's a couple months shy of 6 years old that's also been a daily driver. I've replaced the battery, but it has otherwise stood up surprisingly well. I was amazed when I looked up how old it is.
I'll pass it onto the kids once I get one of these shiny new ones and will be replacing the keyboard and fan when I do. Some of the plastic covering on a few of the keys is starting to come off and the fan bearings are starting to go so sometimes it's louder than it ought to be.
