Other manufacturers have recalls too, for equally dangerous conditions, but only Tesla ever makes the news (I'm not defending Tesla, just proving a point). CT has been out for ~4 months, 4k units delivered, platform engineered from scratch, new low voltage system, steer by wire, ...... inevitably something will slip.
Compare a brand-new everything to minor updates to existing cars:
> People will buy nice electric cars that aren't riddled with quality issues.
Anecdata: my father bought a VW ID.4 1 year after I got my M3 (2019). The MY wasn't available yet. VW ID.4 has been at the garage 3-4 times/year for various (mostly software) issues. Same with my mother's Peugeot e-208. My M3 got the latest software update the next day after an imperfect firmware update made autopilot sometimes unavailable, all done with the technicians via the Tesla app.
You can choose between amazing efficiency and software, and perhaps a 1mm-off panel on 1 out 1000 vehicles, or perfect esthetics but horrible software.
Teslas are still not perfect, and a stronger QA would definitely make for a better image, I agree. Still, Tesla has made a lot of progress, you don't find as many people complaining about QC/panel issues anymore on reddit. But competitors still have a ton of catching up to do.
While I agree that Musk's personality is damaging Tesla a lot lately, and it lost a huge chunk of the accumulated goodwill, I still don't get how people manage their daily lives then. Shall we stop buying Nestle products? Luxottica? Oil companies? Clothing products? The list goes on.
Besides that, the closing sentence and some comments regarding stalks and buttons demonstrate once again that many people have never driven a Tesla for more than a 2h rental:
> Of course, there is much more competition today, and Tesla’s designs are getting old, and looking a little stale. Their huge decade-long software advantage is now probably somewhere in the 2 -4 year range.
The design is minimal due to aerodynamic reasons, and competitors have essentially the same lines, with perhaps a slightly more rounded accent to it. Interiors are minimal because half the point of Tesla is save on costs on 200 buttons that you never use once. The software makes up for it 99% of the time.
Similar priced cars in the same segment are still nowhere near the efficiency or software quality of Tesla. Yes, FSD is still in beta, sure, but the infotainment experience is still unmatched, and it's not a 2-4 year headstart, it's more in the 5-10 year range.
Yes, but you need to build a similar network of hydrogen pumps as we already have for gas, hydrogen requires more power and infrastructure to store, requires more energy to extract, cars need pressurized tanks, ............
Batteries are simply cheaper, and require far less infrastructure.
"We fucked up the air globally in <50y, lets fuck up the ocean faster to un-fuck the air issue temporarily".
But seriously, we lack the understanding to do local geo-engineering, much less at global scale. Hard pass.
Reducing CO2 emissions by going green and EVs (and not just EV cars), and removing CO2, is a longer but much more controllable and stable approach. I just wish we had better battery technology already, but we spent the last 100y deluding ourselves that a finite non-renewable energy source, that messed up our only biosphere, was the only course of action.
I run kubernetes as... a standalone node. I run it on a dedicated server with two disks on btrfs RAID1, with a subvolume for each pod that needs it (hostPath).
Not the minikube version, not the kubespray version: the kubeadm version, installed by hand and minimally tweaked (NodePorts 0-65535), with dual-stack networking support.
I have ~3 tenants, 5-15 pods each, and managing this configuration with docker was a steadily increasing pain for the last few years:
- I had docker-compose split-files for each service for each tenant to keep a consistent state, so I used a simple bash script to generate a really long docker-compose line to bring everything together, networking was a real pain
- Dual-stack support in docker does not exist/work in compose v3 to this day, but v2 has it
- Querying the configuration and state of a container was an exercise in debugging the output
- For certificates, I had to run letsencrypt/certbot externally (2 tenants with separate accounts for wildcard certs), but use traefik in docker-compose (the syntax pain was on top of the rest, of course)
Kubernetes simplified all the pain points above:
- I still have the service split-files, but I don't care about network addresses and links between pods anymore, just their names
- better isolation between all the architecture points, and clearer definition and integration
- IPv4&IPv6 native just working
- "describe" gets me all the information I need in a pretty format, "get type name -o yaml" if I need the configuration
- standard nginx ingress + cert-manager, and I can have multiple tenants without resorting to hacks
DBs are regular containers in the pod that requires them, just like any other container. Daily subvolume snapshots are small and effective to make sure the data is safe.
Kubernetes is a much better orchestrator than docker compose, availability is a plus if you really need it. You don't need to build an HA control plane plus 3 workers if you have no use for it.
I work in a small software shop (~10 people), and besides me there's only one other guy that could take up the task of running wires all over the office, the others are developers, not sysadmins. We don't have >3k$ "business" desks, we have Ikea desks. Our office is essentially an apartment converted to office. It does have ethernet wall sockets in each room, but it's only 2 per room.
We've been talking about wiring the office properly for years, but:
- we'd have to run wires on the ground from the wall sockets and protect them properly, otherwise creating potential safety hazards (people tripping, water, ...)
- we'd have to add a switch per room, to place somewhere, but the only available places are the cable-holding Ikea desks (not exactly reliable)
- we'd have to run and hold cable on each desk, for.... people that mostly use Macbooks, most of which only have USB-C ports; we'd have to buy expensive docks just to fix the ethernet cable in place, when all our screens have power-delivery over USB-C already and everyone has wireless keyboard/mouse
To us the investment is simply not worth the effort, not just in money, but also in time spent implementing it and maintaining it. It's simply not a priority, we have applications to deliver.
I did work in banks and big corps, and there I always had wired desktops with wifi being reserved for guests (with certs generated by the host ofc). I can totally see small/medium-sized companies growing organically over the years and not getting this addressed as a top priority, as you said wifi is fast enough nowadays.
- quality has been steadily and consistently improving
- they have the SC network that everyone is adapting to
- they managed to slash prices, reduce per-car costs, while still getting the average auto industry margin - there is no EV competitor that can do this without selling at a loss, at the moment (if they do, everything becomes an optional and you get back to 50k$)
- finally a Non-US-sized vehicle is going to be released in 2025 - definitely 2y too late due to the CyberFuckup (seriously, the only market capable of buying this is the US)
Everyone keeps comparing to BYD, but I just read today that the software quality and compat for charging are simply not there yet. The other manufacturers are even more behind.
What I see/think:
- Wall Street pseudo-science of EPS has been flawed since forever: the same analysts expect Tesla to keep growing >50% YoY, and call for doom if it misses 600mln out of 25bn, or 0.03$ out of 0.71$ (I see this all the time also with other stocks, esp. in the last ~4y)
- Tesla's valuation is heavily reflecting the expected future developments, and has been inflated for some time; this is not necessarily a bad thing, but a market correction was to be expected and long overdue
- Musk's cult and news outlets created this alternate reality where having a Tesla product expresses approval of the CEO: I don't buy it, and I don't care; I own a M3 and plan to get a Powerwall because they're the best products in the price range
- Musk's behavior is concerning, I'd rather see him out at some point, but I fear Tesla would lose its most powerful asset: the constant, enormous push to innovate and break boundaries; if Tesla were to become the regular car company, it would probably disappear in less than 5y
- I hear all the time on Reddit that Service Centers are a hit and miss (esp. in US), this will need to be addressed ASAP not to break user trust (along with repair costs); my experience has been flawless luckily
- their products are way superior to competition for the price range, other car companies cannot get even close to the efficiency and reliability of Tesla vehicles, not to mention the UX/software updates (not FSD)
- FSD 12.1 video released a couple days ago is impressive, I finally get the feeling we're getting close to something usable (I don't have it and will very likely never buy it)
- CyberFuckup should have been canceled before prototype phase, and deposits refunded; it was and still is a massive resource distraction on the way of bringing EVs to the masses
- a 25k$ vehicle should have come sooner, it's coming, and will sell like crazy when available, esp. outside the US
- I hope storage and solar won't be forgotten, all other solutions I've seen so far are not so well integrated with the house, or as easy to control via a phone app
Some anecdata from daily cold booting my MacBook Pro 16' 2019:
- Slack is very reactive, displays a spinner in <2s, takes a few seconds (<10s) to fully load and draw - perfectly acceptable
- Outlook (new experience mode) is a bit slower, takes a few seconds extra to draw, but is still reasonably fast and manageable considering the amount of data it handles - still reasonable, although a bit laggy at times
- New Teams ® (released somewhen Q3-Q4 2023) takes >15s to open or display anything, >15s to draw, >5s to switch across organizations, >1s switching chat tabs in the same org
I'm not spending 3k$ every 2-3 years because "my computer is the problem".
We need to force devs to use and test on non-workstation machines, that don't have 64GB of RAM and the latest CPU/NVMe drive, especially for Electron apps, which open an entire browser copy for each application.
Been driving a M3 since 2019: at this point I feel most of the people complaining about Tesla UX have never really driven one for more than a rental day.
Yes, everything is in the tablet, but it's super-responsive (even on the Atom CPUs from pre-2021 models), essential buttons are well placed, and the car does a good job of adjusting AC (air temp/flow/outlets) and lights automatically. Once you get to know the car, it really does a good job of staying out of your way most of the time.
Compared to my father's ID.4 or my mother's Peugeot e-208, Tesla's software already had a 10y advantage in 2019, and it still has at least that much after 5y. Their car is laggy, menus are too deep to navigate during drive, lots of useless options placed alongside common ones.
I do agree with jnsaff2 that I probably wouldn't like the Highland refresh with no stalks and no USS, I also feel these two things were better kept as they were.
...with which you still need to fight with MongoDB.
I'm still running with a Turris Omnia as router, which served me very well for >7y, added a small legacy SSD, managed to setup everything and worked for a while, but TO is armhf architecture, which mongodb quit packaging for with version 3.x. LXC containers for armhf (Debian/Ubuntu) stopped being released too by official channels. But the latest shiny Unifi app requires MongoDB >= 4.x (Unifi app >=7.5).
"Ok, I'll just buy one of those small embedded boxes": purchase Odroid H3+ (x86_64), install stuff, add docker, start up mongodb5 container annnnddd... illegal instruction! Turns out with MongoDB 5 they decided that with pre-built packages everybody has AVX instructions anyway, and packages are built with the expectation AVX instructions are present... which is not always true for low-power devices or even servers (the Intel Jasper Lake CPU of the Odroid H3+ was released in Q1'21).
In aviation autopilot nowadays can do much more than just "keeping the plane straight", it can also improve stability during maneuvers, or auto-land (mostly, ILS). But even aviation autopilot has limitations, which are well understood by the well-trained officers operating the plane as their only mission.
I see these problems with the naming:
- non-technical people have a "magic" grasp of the "autopilot" word that was passed down by movies, as a "set-and-forget" magical button that will make all the problems go away
- non-technical people won't read manuals (most of them, at least), so they are not aware of limitations, and WILL get distracted thinking AP will safely operate the car for them
- Tesla's AP, as of today does NOT handle most streets in most conditions well enough, and still has some caveats on highways, which are its intended operational domain; I understand FSD would do better, but still not a set-and-forget experience
So while I agree that it's being undersold feature-wise, I really think the issue is that "autopilot" is tied to unrealistic expectations in the mind of the average person, which easily leads to dangerous situations in the case of operating a 2-ton vehicle.
If people are worse on average, on average they'll cause more accidents.
They won't get a firmware update, but they'll get an IM while they're late, won't pull over to reply, and ran you over at full speed instead of trying to slow down.
You pay to have traffic visualization, but traffic data is part of the trip planning, and the navigator does reroute you on a faster road if it's available (the delta is configurable).
> It wasn't the full Autopilot, just the lane following/adaptive cruise control kind of thing but we almost did go off the road at least once.
Your friend should have warned you that there are conditions where AP is expected not to work well, they're also clearly stated in the manual. I tend to agree that Tesla oversells the AP/FSD functionality just by the name, AP should be renamed to "vision lane assist", or something along those lines.
> But in general with the UI, there was a lot of, "Hey, how do I do [elementary car function]" and he'd be like "Oh I think they changed that [or moved it in the menu] in the last update, uh let me have a look"...
As an M3 2019 owner, I recall 2 instances since 2019 where stuff was moved around a lot, otherwise it's pretty stable.
While the first big update sacrificed some navigator real estate space for non-FSD users as well, it also grouped speed, speed limit and AP limit better at the top left corner, which used to be placed further away (now I don't need to turn my head anymore, I just glance bottom-right). The second one introduced dynamic tiles at the bottom for the less used stuff (things which were otherwise hidden behind non-intuitive areas, like the T Tesla logo at the top center of the screen).
After 6~10mo ago (can't recall) they grouped all car commands in the same panel when you bring up the settings to make it easier and more consistent to access those when required.
UI has gotten much snappier since I got the car, even though it's still running on a very low-end Intel Atom chip.
My feeling is that some breakage is sometimes required to achieve better solutions, and while there's the usual outrage of "changing my workflow" ® in the beginning, Tesla UI updates are usually for the better in the long run.
Perhaps it's because I work in a small software shop and we do only B2B, but 99% of our applications consist of a frontend (JS served by an nginx image), a middleware (RoR, C#, Rust), nginx ingress and cert-manager. Sometimes we have PersistentVolumes, for 1 project we have CronJobs. SQL DBs are provisioned via the cloud provider. We monitor via Grafana Cloud, and haven't felt the need for more complex tools yet (yes, we're about to deploy NetworkPolicies and perform other small changes to harden the setup a bit).
In my experience:
- AKS is the simplest to update: select "update cluster and nodes", click ok, wait ~15m (though I will always remember vividly the health probe path change for LBs in 1.24 - perhaps a giant red banner would have been a good idea in this case)
- EKS requires you to manually perform all the steps AKS does for you, but it's still reasonably easy
- All of this can be easily scripted
I totally agree with the other comments here: LTS releases would doom the project to support >10y-old releases just because managers want to "create value", but don't want to spend a couple weeks a year to care for the stuff they use in production. Having reasonably up-to-date, maintainable infrastructure IS value to the business.
Not having to constantly apply a ton of micro-corrections for keeping the speed, keeping the car centered, slowing down to cars in front, ... takes away a TON of mental fatigue, even if you're still supervising.
Over time you learn the 1-2 situations where the car potentially messes up and you increase vigilance, otherwise highways are usually a smooth ride.
The first time I took the trip to my parents in my Tesla, I almost did the entire trip without stopping (~300km), whereas with my old car, which had no assistance systems at all, I would have to stop after 1h from being too tired already (Switzerland has lots of speed limit changes and frequent traffic jams near the weekends).
I honestly couldn't tell too much difference between my M3 1-pedal driving (stop mode: Stop) and my father's ID.4 (not sure if it was in B mode). I used to go with Creep mode (Stop mode was introduced sometime in 2020 IIRC), then I tried Stop and fell in love with it, it provides max regen and auto-engages breaks when close to standstill.
As for the overspeed, on highways, I usually ensure there's enough space in the passing lane, I change lane and let it accelerate without pushing it myself (or give it a small nudge without disengaging). I literally just discovered that AP also has "Overtake acceleration", which I never knew existed or used ( https://www.tesla.com/ownersmanual/2017_2023_model3/en_eu/GU... ). On city streets, I let TACC run at 50km/h or 30km/h, so I usually go for the accelerator only to navigate in/out of turns.
You can switch on TACC at <30km/h if you have visible traffic ahead of you more than 1.5m apart, so unless you're driving at night with nobody around you should be able to enable it also at 20km/h.
TBH though after 4y I also developed a feeling for city-speeds, so I rarely overspeed, even accidentally. The small price to pay is that I really have to use feather-weight on the accelerator and it's not always easy (esp. if you're physically tired).
As I read your post I realized what you really want: power cut-off at set speed ("do not push over X") with override option. Throttle remapping sounds cool, but I'm not sure anybody implements it like that. AFAIK Tesla does not have that on any model, agreed.
Replying by point, as I've been driving my 2019 M3 for 4y now:
- agree, rent it for a few days after the test drive, EVs have a different style of driving that might not suit you
- pings in modern cars are everywhere, it's not just Teslas - the way I drive, I very rarely (<2 times/month) get a ping from Forward Collision Warnings, but just because I keep it to "Early", which is the longest triggering distance; otherwise the car is silent (you may be constantly driving too close to other cars?), my father's ID.4 is just as annoying, if not worse
- people call Tesla interior "cheap", when in reality it's just simple and plain but durable, most of the costs of the car are in the battery and powertrain
- Am 1.85m tall at 110kg, steering wheel feels appropriate in size
- On my M3 2019, there's two buttons that have 2 axes (up/down/left/right) and a push
- Overspeed indication can be configured: none/speed limit icon animation/super-annoying bell that scares you every time (I keep it to icon animation)
And, replying to final point, the speed limiter IS the adaptive cruise control, which is perfectly working in cities, esp. in traffic jams. The way it works is to pick the speed limit as ceiling, and adapt the cruise to the traffic and the follow the vehicle in front at the distance that you've set (right wheel button left/right). The distance is adaptive (dist. 7 @ 50 km/h << dist. 7 @ 130 km/h). I usually keep 7 because it reduces the amount of slowdowns when people cut in front, resulting in a smoother ride. I've often seen the complaint that follow distance 7 in US cities is impossible, because everyone would jump in front of you - I'd still not recommend setting 2, it makes for a super-annoying stop-and-go drive.
I have the "bad luck" of living in Switzerland, where speed limits change every 50m: my routine is to use TACC as much as possible, and disengaging it with ample margin (using only regen to break) when coming up to traffic lights/roundabouts/crossings. I don't have to think about speed limits anymore, because TACC does it for me. Only exception is highway drives, where I need to change the speed limit beforehand with the right steering wheel button (or tap on the limit icon).
Hint: you can disengage TACC and/or AP without breaking, just do a quick soft tap-up on the right stalk (like soft-blinker tap).
There are a couple important points about the driving style of an EV, that I feel are not stressed enough:
- you try to regen as much as possible and avoid using breaks to get some energy back (extend the driving range), and you wear down break pads way less than ICE cars
- you don't floor it after every stop, because torque is instant and the engine needs no revving up - if you do this, I guarantee you'll always be running up to the bumper of the vehicle in front of you and having the car scream at you (regardless of the brand)
From my point of view, after 4y, looking also at all the other EVs that followed after Teslas (father has ID.4, mother Peugeot e-208), Tesla has managed to automate mostly-well many aspects of driving while reducing UI complexity (and removing "I have used it once in the lifetime of the car" buttons).
- AC does a good job of keeping the temperature in relation to the outside temp, also selecting the appropriate air flow outlets
- Automated lights off/on/high-beams well enough, that honestly I can't remember the last time I touched the setting, possibly never
- Voice control works decently enough that I can select songs or playlists just by pressing the wheel button (in addition to controlling pretty much all the car features)
- Auto-wipers works well enough for me in >90% conditions, except perhaps hurricane-level thunderstorms in a pitch-black area
- TACC/AP do an excellent job at keeping distance and lane on highways, and they're years better in terms of quality compared to competitors (I wanted for a long time Enhanced AP with lane changes, but after having tested it, I'd rather not; never used FSD, also: Europe)
Considerations about the UI: people keep claiming that they want 200 physical buttons, and that it's dangerous to take your eyes off the road, but:
- the Tesla UI is done well enough that I can adjust my AC faster than my ICE-powered friend can (not that I need it that much, it does a good job)
- my friend still also needs to either look at the AC temp/flow selectors or press a -/+/... button until a display shows what he wants, so it's not like the ICE version is safer than the Tesla in this regard
- I feel the "muscle memory" everybody talks about works for 1/2 HUGE tactible features at most if you only have 1 car available to drive, the rest of stuff you'll still need to look what you're doing regardless (unless you regularly spend >15s counting X buttons from the left of the second row of buttons from the bottom)
- TACC and AP (and safety features worst case) fill in the attention void created by these actions, while the ICEs have none of these available unless they are in the same price range
- VW ID.4 and Peugeot e-208 have archaic OSes that have huge input latency and were designed with automotive in mind BUT NOT REALLY TESTED FOR IT (in general UI interactions are NOT well thought out), while Tesla has a simple settings panel with everything in it, automates most regulations, and it's super-responsive
- VW ID.4 has so many buttons to activate the TACC/AP equivalents, in different modes, that my father after >2y of daily driving still hasn't mastered them, and the steering wheel looks like an Formula 1 steering wheel; compare that to: "single-tap" -> TACC engaged + "double-tap" -> AP engaged
In general, driving for me has become fun again: I don't need to worry about a ton of stuff, and operating the car is WAY easier than manual or even auto transmission.
> Put another way, if you look at all the features that I want in a car (robust safety cell, powerful propulsion, quiet, smooth ride, rare) none of them can be upgraded without physical modifications. And in fact, once those things are achieved, there is really very little reason to change anything.
disclaimer: I own a 2019 M3.
IIRC Tesla's acceleration boost (powerful propulsion) upgrade essentially keeps the front engine correctly-aligned at all times to make the most out of the usually-unused torque. AEB and TACC were upgraded multiple times over the years, becoming better and better and solving issues (smooth ride). As for quietness, MS and MX got active noise reduction through software upgrades.
All of this was done using existing hardware, and I didn't even mention the constant data-driven tweaking of battery pre-conditioning to allow faster charges while still extending battery life (powerful propulsion/smooth ride), or the camera features (dashcam recording, blind-spot camera pop-up, ...).
Some are amazing tweaks you didn't even know you wanted, but are a constant time-saver (blinkers auto-off).
I'm pretty sure that also the other high-voltage components handling also got the data-driven treatment (better regulation of currents to extend component's lifetime, for instance).
These kind of software modifications were also seldom done when I brought my old ICE cars to the dealer for the regular checks - "remapping the engine parameters to better handle some cases" - but in reality I recall this happening at most once in the lifetime of the car.
I believe that new features can be developed using existing hardware in new, unforeseen ways. What I agree on is that this needs to be done in a secure and non-intrusive manner. Some UI reworks were definitely controversial at first, but Tesla has a good track record of listening and adapting to user requests at least.
I cannot say the same for any other brand I've had a car from.
Compare a brand-new everything to minor updates to existing cars:
- https://www.cbsnews.com/news/toyota-lexus-recall-2020-2022-f...
- https://www.cbsnews.com/news/toyota-recall-50000-corolla-rav...
- https://www.cbsnews.com/news/toyota-recall-rav4-suv-2023/
- https://www.cbsnews.com/news/toyota-recall-tundra-fire-risk/
And this is just Toyota, which as far as I can gather is considered reliable and trustworthy.