Enhanced Summon: slow-mo driving in parking lots

[S4WRXTTCS]

Wow, exactly a week later. What was he thinking would happen within one single week? How can something like this happen?
You would think if something is about to be released in a week, the only thing missing would be some finishing touches.

I mean sure, Elon is a pretty optimistic guy, I get that and making estimations over several months, or years is hard. But this was a single damn week! Did they accidentally delete their code?

I think there are two plausible possibilities.

1.) It's clear from the early previews of Enhanced summons that it was relying on some really bare bones navigation planning. To me it's entirely possible that they had more sophisticated versions that weren't part of the mainline. Something that a few coders were working on in hopes of switching it to that. Or maybe it was unacceptable pauses that they were trying to remove, but ran out of time. In any case it's not unusual for software teams to fail to deliver. It's also not unusual to simply say "this isn't where it needs to be", and put it back in the oven.

2.) Elon simply didn't want to time the release of Enhanced Summons so close to Autonomy day on the 22nd. It doesn't make much sense to demonstrate the next-gen capabilities so close to what's likely going to be a disappointment (due to unrealistic expectations).

Which one is true? It entirely depends on how optimistic/cynical one chooses to view Tesla/Elon.

 

[diplomat33]

I think there are two plausible possibilities.

1.) It's clear from the early previews of Enhanced summons that it was relying on some really bare bones navigation planning. To me it's entirely possible that they had more sophisticated versions that weren't part of the mainline. Something that a few coders were working on in hopes of switching it to that. Or maybe it was unacceptable pauses that they were trying to remove, but ran out of time. In any case it's not unusual for software teams to fail to deliver. It's also not unusual to simply say "this isn't where it needs to be", and put it back in the oven.

2.) Elon simply didn't want to time the release of Enhanced Summons so close to Autonomy day on the 22nd. It doesn't make much sense to demonstrate the next-gen capabilities so close to what's likely going to be a disappointment (due to unrealistic expectations).

Which one is true? It entirely depends on how optimistic/cynical one chooses to view Tesla/Elon.

Option 1 seems pretty similar to my explanation. I also don't think 1 and 2 are mutually exclusive. There might be some truth to both. I definitely see your point that a lackluster ES release before the April 22 event might have tainted the event a bit. With the delay, they can hopefully have an exciting April 22 event and then release an even better ES soon afterwards. That would certainly make for better PR.

 

[R.S]

I think there are two plausible possibilities.

1.) It's clear from the early previews of Enhanced summons that it was relying on some really bare bones navigation planning. To me it's entirely possible that they had more sophisticated versions that weren't part of the mainline. Something that a few coders were working on in hopes of switching it to that. Or maybe it was unacceptable pauses that they were trying to remove, but ran out of time. In any case it's not unusual for software teams to fail to deliver. It's also not unusual to simply say "this isn't where it needs to be", and put it back in the oven.

2.) Elon simply didn't want to time the release of Enhanced Summons so close to Autonomy day on the 22nd. It doesn't make much sense to demonstrate the next-gen capabilities so close to what's likely going to be a disappointment (due to unrealistic expectations).

Which one is true? It entirely depends on how optimistic/cynical one chooses to view Tesla/Elon.

Over the years that I've followed Tesla, I've learned to take the announcements with a grain of salt. In the case of FSD even with a bag of salt. But what angers me the most, is how it's always the same cycle of announcing something and then delaying something. And this time it even was within one week.

Sure, things can get delayed and maybe there is a very good reason for all of it. But why does he keep announcing stuff like that? It's not even like Tesla gains something from him announcing ES.

And please Elon, don't show us another FSD demo video on April 28th. We have already seen it.

 

[S4WRXTTCS]

Over the years that I've followed Tesla, I've learned to take the announcements with a grain of salt. In the case of FSD even with a bag of salt. But what angers me the most, is how it's always the same cycle of announcing something and then delaying something. And this time it even was within one week.

Sure, things can get delayed and maybe there is a very good reason for all of it. But why does he keep announcing stuff like that? It's not even like Tesla gains something from him announcing ES.

And please Elon, don't show us another FSD demo video on April 28th. We have already seen it.

I do find it interesting how people react to his tweets.

I reacted positively to the "it's not quite ready" tweet of Enhanced Summons. I can accept that if needs a little more time than it needs a little more time. I don't see any reason to rush it.

I reacted hugely negatively to the "FSD will rapidly increase in price".

Why the different reactions? One is a known deliverable with what I expect to be an incremental improvement over what I've already seen from the videos from early access people. I could see reasons for delaying it from the videos I've watched. It's also extremely important to give people updates on the progress of features, and bug fixes. Tesla owners need to be given status updates on features they paid for. I expect Tesla/Elon to get theses wrong because estimated are hard, but they're also necessary.

The other one largely feels like fear mongering. Why even tell people the price is going to go WAY up? The only reason I can think of is to try to get people to BUY now on something we haven't even seen in action.

What investors have been promised for April 22nd (last I heard it was 22nd) is actual test drives.

They also can't fake any of it because the FSD video from 2016 was done using Nvidia SW (the stuff Nvidia delivers with the Drive PX2 computer). They can't run that software on HW3.

90% of the reason to be excited for the 22nd is Karpathy, and not Musky.

 

[R.S]

I do find it interesting how people react to his tweets.

I reacted positively to the "it's not quite ready" tweet of Enhanced Summons. I can accept that if needs a little more time than it needs a little more time. I don't see any reason to rush it.

I reacted hugely negatively to the "FSD will rapidly increase in price".

Why the different reactions? One is a known deliverable with what I expect to be an incremental improvement over what I've already seen from the videos from early access people. I could see reasons for delaying it from the videos I've watched. It's also extremely important to give people updates on the progress of features, and bug fixes. Tesla owners need to be given status updates on features they paid for. I expect Tesla/Elon to get theses wrong because estimated are hard, but they're also necessary.

The other one largely feels like fear mongering. Why even tell people the price is going to go WAY up? The only reason I can think of is to try to get people to BUY now on something we haven't even seen in action.

What investors have been promised for April 22nd (last I heard it was 22nd) is actual test drives.

They also can't fake any of it because the FSD video from 2016 was done using Nvidia SW (the stuff Nvidia delivers with the Drive PX2 computer). They can't run that software on HW3.

90% of the reason to be excited for the 22nd is Karpathy, and not Musky.

Oh yea, it's the 22nd, not the 28th as I said. And I agree the announcement of the "imminent price increase" is just a sales tactic. And while it isn't really cool to employ sales tactics like that. There is some calculation, reason and planning behind them and I can respect that.

When I see posts like the one about ES, it's just endless facepalms. It's loosing credibility for the wrong reasons. Rather than for the right reasons (cash).

 

[diplomat33]

Sure, things can get delayed and maybe there is a very good reason for all of it. But why does he keep announcing stuff like that? It's not even like Tesla gains something from him announcing ES.

Probably because he wants to keep his twitter followers and Tesla customers up to date, especially when it comes to new features like ES that he is very excited about and that he knows we are excited about as well. So when the dev team decided "ok. let's release ES next week", he passed that update along to us. And when the dev team decided to postpone, he passed that along to us as well.

 

[diplomat33]

The other one largely feels like fear mongering. Why even tell people the price is going to go WAY up? The only reason I can think of is to try to get people to BUY now on something we haven't even seen in action.

Well, but isn't it a good idea to give your customers a heads-up before you do a big price increase? If Tesla had said nothing, I guarantee you that customers would be getting the pitchforks out if the price suddenly jumped to $8000 on May 1 with no announcement. That type of stealth increase in pricing would be widely condemned and rightfully so.

90% of the reason to be excited for the 22nd is Karpathy, and not Musky.

Agree 100%. As I said before in another thread, it's really Karpathy's FSD, not Musk's FSD. It's Karpathy who is in charge and in some cases directly writing the software. And it's Karpathy who single-handedly saved AP2 IMO. I am convinced we wouldn't be looking forward to an April 22 event if it wasn't for Karpathy.

 

[R.S]

Probably because he wants to keep his twitter followers and Tesla customers up to date, especially when it comes to new features like ES that he is very excited about and that he knows we are excited about as well. So when the dev team decided "ok. let's release ES next week", he passed that update along to us. And when the dev team decided to postpone, he passed that along to us as well.

So how much information did those two tweets actually have for us? Close to 0? And what did Tesla gain from them? Also nothing? Then why even tweet them?

Jobs made Apple a cult brand, by giving out information only at certain events and when he did, he delivered. And people went crazy.

This is how it’s done. Not write some tweet about something, then delay it, then tweet about it again, then delay it again and so forth. Imagine he would have never talked about FSD and then just revealed NOA. At a big event. People would have gone crazy. But since he has teased so much more over the years, NOA is cool, but not „the real deal“.

 

[diplomat33]

So how much information did those two tweets actually have for us? Close to 0? And what did Tesla gain from them? Also nothing? Then why even tweet them?

Jobs made Apple a cult brand, by giving out information only at certain events and when he did, he delivered. And people went crazy.

This is how it’s done. Not write some tweet about something, then delay it, then tweet about it again, then delay it again and so forth. Imagine he would have never talked about FSD and then just revealed NOA. At a big event. People would have gone crazy. But since he has teased so much more over the years, NOA is cool, but not „the real deal“.

I agree that would be a much better marketing approach.

 

[iTech]

I agree that would be a much better marketing approach.

Looks like Musk is disrupting the marketing business as well!

 

[banned-66611]

Enhanced Summon delayed "a few weeks" according to Musk on Twitter:

View attachment 396589

Chilling to think that they will release it onto public roads mere days after they deem it "good enough". Testing such a feature should be many months at the least.

 

[diplomat33]

Chilling to think that they will release it onto public roads mere days after they deem it "good enough". Testing such a feature should be many months at the least.

Enhanced Summon is not used on public roads. Second, Enhanced Summon has been in testing for several months already.

 

[banned-66611]

Enhanced Summon is not used on public roads. Second, Enhanced Summon has been in testing for several months already.

It's used in public car parks. If anything that's worse, because car parks are places where cars and people mix.

 

[diplomat33]

It's used in public car parks. If anything that's worse, because car parks are places where cars and people mix.

Yes and? The last videos we got of Enhanced Summon looked pretty good actually. Enhanced Summon seemed to be working pretty well. It moves fairly slowly too so there is little risk to people, especially since the owner has to monitor and keep their finger on the button for it to work. It looked ready for public release so the fact that they delayed it is a good sign. It means we will get something even better.

 

[electronblue]

Well, if you look at Musk's last tweet about delaying the release, he does not say that they ran into last minute issues or that they need to put some finishing touches on it. He says "not good enough yet" which I interpret as saying "Enhanced Summon is not good enough for release". But how could ES be ready to roll out and then a week later "not be good enough"? The only explanation is that they did truly plan on releasing ES this week but changed their minds and basically went "we can do better, let's keep working on it more". Imagine a student who studies hard enough to get a B on a test and is content with that effort but then decides to study some more and go for an A. That's how I see it. ES was ready to roll out this week but the team decided at the last minute to work on it some more to make even better. Maybe at the last minute someone on the team had an idea for making ES better so they decided to delay release until they could implement the new improvement? Or maybe with the upcoming April 22 event, the team decided to push harder and make ES even better for the event? Now maybe Musk is just putting a nice spin on it and the reality is that ES is having issues but that's my reading of the two tweets.

Ultimately, these delays happen all the time in software development and having the team work on ES more will only mean a better product for us. So patience is a good thing. We'll get ES eventually and it will be better for the extra work that was put into it.

The problem in the meanwhile is Elon keeps handing out there timelines that just don’t pan out. Perhaps it is for PR effect and/or perhaps it is to buy time. Either way by now he really should know better.

You don’t go constantly from a release in just a very short timeframe to delays of weeks or even months (or years in some cases) unless there something wrong with the way you decide on the first timeline in the first place.

 

[malcolm]

Agree 100%. As I said before in another thread, it's really Karpathy's FSD, not Musk's FSD.

Why would it be Musk's? You could equally argue that sales of Teslas would be through the floor if it wasn't for von Holzhausen and invent a narrative where Franz is constantly saving Elon's a$$.

But are we getting to that stage where, if we love the Tesla pickup design, we'll praise Franz (correctly), or if we love APs Advanced Summon, we'll praise Andrej (correctly) and if we don't we'll blame Elon for either/both?

I guess that's just "business" rather than logic.

I agree that Musk's twitter account would benefit from more / any "Don't thank me, thank...." - styled tweets. Wouldn't hurt.

Elon does push himself and others to produce their best work within short time-frames. By all accounts he is difficult to work with/for and there are negatives to that (executive turnover, over-promises, tweeting-while-high) as well as positives (ground-breaking vehicles that every industry player is working to overtake).

How people on this forum feel about all of this, largely depends on whether their own money is/has been involved and whether they feel that the value of that money has been honored professionally and in a reasonable time-frame. As always YMMV.

 

[DrDabbles]

The big things that these videos proved to me the instant they were published:

1. Autopilot doesn't divide a surface into halves for bi-directional traffic, or care about left vs. right side driving.
2. Autopilot doesn't interpret most markings on road surfaces- parking lot lines used to indicate driving vs. parking area for example.
3. NoAP is truly a GPS assisted line follow robot, which explains why it's terrible at figuring out what lane to choose at splits, merges, and exits.

I sincerely hope, since HW3 has existed for around a year at least (after development), that a decent software stack has been developed and it's able to just do enough raw processing that it's fairly reliable. Current AP is okay in several situations as long as you're in the middle lane(s), so improvements to that would be a fairly significant change. Just, not when compared to the marketing and rabid fans.

 

[electronblue]

The big things that these videos proved to me the instant they were published:

1. Autopilot doesn't divide a surface into halves for bi-directional traffic, or care about left vs. right side driving.
2. Autopilot doesn't interpret most markings on road surfaces- parking lot lines used to indicate driving vs. parking area for example.
3. NoAP is truly a GPS assisted line follow robot, which explains why it's terrible at figuring out what lane to choose at splits, merges, and exits.

I sincerely hope, since HW3 has existed for around a year at least (after development), that a decent software stack has been developed and it's able to just do enough raw processing that it's fairly reliable. Current AP is okay in several situations as long as you're in the middle lane(s), so improvements to that would be a fairly significant change. Just, not when compared to the marketing and rabid fans.

Indeed the currently very limited (and even then rather unreliable) ability of Tesla Vision (what they called the vision part of AP) to identify road markings and obstacles is something that sets it apart from Waymo and MobilEye that have the ability to, not only do it more reliably, but recognize more of different types of elements on the road — be it marking, signs, different levels and types of curb etc. Based on publicly seen information of course, what HW3 and a new NN might unveil remains to be seen.