Can we see more reviews like this one?

Him being an MTB expert doesn't make him an ebike expert... Hard to understand it?

He misses a long time experience in e-bike systems; simply because he cannot afford e-bikes or has no sponsors to provide e-bikes to him. The Original Post is the proof.
Im pretty sure he can afford one, have you seen his view count.
That one video probably earned him 60K
 
The tread, middle, edge side and a nice tan sidewa..oh dear its died under explanation :)
Ah, I see now! These Specialized Pathfinder Pros are pretty popular on bikes like mine, but that narrow center slick has traction issues in emergency stops on smooth pavement and on steep offroad descents. But being a brittle old guy with spinal hardware, I'm kind of a chicken on the latter.
 
Unless you are just trying to make it complicated, it is hard to see how normal mountain biking expertise does not translate to riding an e-mtb well. Bluntly I am skeptical that any of the subtleties wouldn't be quickly picked up by a good mountain biker, probably in the first half hour of riding the bike.
 
Ironic, since I addressed that. Obviously, no point in reading anything on this forum, since none of us are experts.🙄
I guess we both know more about e-bikes than Seth would ever know :)

Unless you are just trying to make it complicated, it is hard to see how normal mountain biking expertise does not translate to riding an e-mtb well. Bluntly I am skeptical that any of the subtleties wouldn't be quickly picked up by a good mountain biker, probably in the first half hour of riding the bike.
It is not about riding. How do you describe a guy whose first action post buying an e-bike is opening the battery?
My description is "a clueless amateur".
 
I guess we both know more about e-bikes than Seth would ever know :)


It is not about riding. How do you describe a guy whose first action post buying an e-bike is opening the battery?
My description is "a clueless amateur".
Maybe... Maybe not. Seth's a big Specialized guy and his wife rides a Vado.
 
Maybe... Maybe not. Seth's a big Specialized guy and his wife rides a Vado.
For me that's proof enough that he's a 🤣😂🤣😂🤣😂🤣😂 🙃

And yet others never get tired of. . .
tumblr_mm8gulQvoz1s3g3ago1_500.gif
 
Hes making youtube content, its his only income, its not real life.
He does seem very knowledgeable, but its youtube, he will have researched it, wrote a script, edited and reshot it several times.
Try asking chatgpt to write you a youtube suitable script for discussing various torque and power curves for a midrive.
Its unbelieavable with just the simplest of instruction it can produce a perfectly feasable script at various levels of complexity, this was produced instantaneously.

You said:​

A youtube script for tourque sensing and voltage curves for ebike mid drive motors

ChatGPT said:​

Certainly! Here's a YouTube script for explaining torque sensing and voltage curves for eBike mid-drive motors:

[Intro Scene]
[Upbeat background music]
Host on screen (smiling and excited):

"Hey everyone, welcome back to the channel! If you're into electric bikes, e-bike technology, or just curious about how mid-drive motors work, you're in the right place. Today, we're diving into torque sensing and voltage curves, which are key concepts that affect your eBike's performance. Let's break it all down!"
[Cut to close-up of eBike mid-drive motor]
Host (voiceover):

"When it comes to mid-drive motors on eBikes, understanding how power is delivered to the wheels is crucial for optimizing your ride. Two big factors come into play: torque sensing and voltage curves. But what exactly do these terms mean? And how do they affect your riding experience?"

[Scene 2: Torque Sensing Explanation]
[Visual of eBike showing the motor, pedals, and torque sensor components]
Host on screen (pointing at motor and sensors):

"First up, torque sensing! Torque sensing refers to how the bike’s system measures the force you're putting on the pedals. Essentially, it helps the motor figure out how much power to apply to match your pedaling effort. This makes for a smoother, more responsive ride."
[Cut to animation showing torque sensor data being sent to the motor]
Host (voiceover):

"The torque sensor detects the amount of force you're applying through the pedals. It sends this data to the motor controller, which then adjusts the motor’s power output. More force from your legs means more power from the motor, giving you a natural feel as if you’re riding a traditional bike, but with an extra boost."

[Scene 3: Voltage Curves Explanation]
[Visual of voltage curve graph, with voltage on the y-axis and motor speed on the x-axis]
Host on screen (holding a tablet showing a voltage curve graph):

"Now, let's talk about voltage curves. Voltage is the key to controlling the motor’s speed and torque. The voltage curve helps us understand how the motor behaves at different speeds, especially in relation to the input from the throttle or pedal assist."
[Cut to graph showing a rising voltage curve]
Host (voiceover):

"As the bike accelerates, the voltage increases, which allows the motor to spin faster. But what's interesting here is that mid-drive motors often have a unique voltage curve compared to hub motors. They provide more consistent power as the bike reaches higher speeds, resulting in smoother acceleration and less power drop-off at higher gears."
[Visual transition to an eBike riding up a hill, showing the motor in action]
Host (voiceover):

"When you're climbing a hill, for instance, the torque sensor detects the extra force you're putting into the pedals. The motor compensates by adjusting the voltage and providing additional power, making the climb much easier."

[Scene 4: Torque and Voltage Curve Interplay]
[Visual of torque vs. speed graph]
Host on screen (gesturing to the graph):

"Torque and voltage are closely related. The more torque you apply to the pedals, the more voltage the system will send to the motor, ensuring that you get the necessary power boost. At low speeds, the torque response is critical for getting you started smoothly, while at higher speeds, the voltage takes over to maintain performance."
[Cut to slow-motion footage of an eBike accelerating smoothly]
Host (voiceover):

"Together, torque sensing and voltage curves create a dynamic and efficient riding experience, adjusting to your pedaling style and road conditions in real time."

[Scene 5: Benefits of Torque Sensing and Voltage Curves for Riders]
[Host on screen, smiling]:

"So, why does all this matter to you as a rider? Well, if you have an eBike with torque sensing, you’ll enjoy a much more natural and intuitive ride, especially when it comes to things like pedal responsiveness and hill climbing. And when you combine that with a well-tuned voltage curve, you get smoother acceleration and better power management."
[Visual of eBike rider gliding through different terrains smoothly]
Host (voiceover):

"This means you'll have more control over your bike, less jerky acceleration, and ultimately, a more enjoyable riding experience, whether you're cruising on flat roads or tackling tough inclines."

[Outro Scene]
Host back on screen (excited):

"That’s it for today’s deep dive into torque sensing and voltage curves for mid-drive motors. I hope this clears up some of the technical aspects of eBike performance! If you liked this video, don’t forget to hit that like button and subscribe for more eBike content. Got any questions? Drop them in the comments below—I'm happy to help!"
[Host waves goodbye]
"Thanks for watching, and happy riding!"
[Upbeat music fades out]

End Screen
[Text: "Subscribe for more eBike content!"]
[Links to other videos]

Simplistic but you can dial in the detail reqd.
What a world we are walking into.
 
It is not about riding. How do you describe a guy whose first action post buying an e-bike is opening the battery?
My description is "a clueless amateur".
I think given all of the quality problems with e-bike batteries and the ugly history there that is a reasonable thing to tear down.

I personally would like to see side-by-side teardowns between quality e-bikes and cheap Chinese deathtraps. In particular I'd like to see disassembly of the electronics and the battery. But also they should show if the brakes, drivetrain, and other parts are appropriate for a heavier pedal assist bike. We all know a lot of manufacturers who should know better take shortcuts on components.
 
It was merely an egg sample.
We have scrapped the entire knowledge of humanity and put it in a language model that is gaining exponential ' intelligence' individual knowledge and experience will be irrelevent within ten years
 
It was merely an egg sample.
We have scrapped the entire knowledge of humanity and put it in a language model that is gaining exponential ' intelligence' individual knowledge and experience will be irrelevent within ten years
You really are surrendering to the hype machine there. You ought to keep in mind that the loudest advocates for this future are also trying to jack their share price and make a lot of money.

Seriously, we are taking just the first baby steps here. And given the current directions baby is walking I am doubtful any kind of artificial general intelligence or artificial super intelligence is possible or practical no matter how much time or money is thrown at the problem. That doesn't mean there won't be useful and valuable applications and large fortunes made. But our Machine Overlords haven't yet made their appearance.

On those practical applications, right now very few people are even looking in the right places.

For that Artificial Super Intelligence to happen, a large number of other pieces need to be invented. And no, we don't know how many pieces or what they even are. So I suspect it is going to take a while.
 
Last edited:
I think given all of the quality problems with e-bike batteries and the ugly history there that is a reasonable thing to tear down.

I personally would like to see side-by-side teardowns between quality e-bikes and cheap Chinese deathtraps. In particular I'd like to see disassembly of the electronics and the battery. But also they should show if the brakes, drivetrain, and other parts are appropriate for a heavier pedal assist bike. We all know a lot of manufacturers who should know better take shortcuts on components.
I'd like to see him tearing down a battery on his brand new Levo :) He is setting a bad example as the public would now think it is OK to dismantle e-bike batteries. Well, I've got bored with discussing Berm Peak.

Another example of "clueless amateurs" are those in the government of Poland responsible for e-bike subsidies. Here are several rules to get a subsidy:
  • The minimum battery capacity shall be 360 Wh. (A Vado SL has a 320 Wh battery).
  • The system voltage shall be 36 V. (The Euro law allows up 48 V; Vado SL is a 48 V e-bike).
  • The e-bike shall be made in the EU. (Vado SL has a German motor but it is made in Taiwan). Ironically, Bafang motors are allowed as the Chinese brand was as smart as to build the motor assembly lines in Wrocław, Poland.
Should I say more?
 
You really are surrendering to the hype machine there. You ought to keep in mind that the loudest advocates for this future are also trying to jack their share price and make a lot of money.

Seriously, we are taking just the first baby steps here. And given the current directions baby is walking I am doubtful any kind of artificial general intelligence or artificial super intelligence is possible or practical no matter how much time or money is thrown at the problem. That doesn't mean there won't be useful and valuable applications and large fortunes made. But our Machine Overlords haven't yet made their appearance.

On those practical applications, right now very few people are even looking in the right places.

For that Artificial Super Intelligence to happen, a large number of other pieces need to be invented. And no, we don't know how many pieces or what they even are. So I suspect it is going to take a while.
It is grossly overhyped, but in reality it is beavering away to replace everything.
Once a critical mass of systems communicate on an accepted standard, then it will propogate exponentially.
The US military is already looking to protect nuclear weapon systems from it.
They are pushing ahead with their own admission of 'see what happens', already passwords, home addresses, sensitive documents, entire critical databases and security diagrams have been revealed from simple enquiries.

At the moment its just an epic search engine that humanises the results, but
The actual king of AI is named after the glass seeing balls from LOTR.
 
Once a critical mass of systems communicate on an accepted standard, then it will propogate exponentially.

That's what people seem to want to believe, for sure. For myself I don't see really credible evidence of emergent behavior from current AIs. Certainly not on the same complexity levels that we'd expect from humans, or even bumblebees. Seriously, you can't model bee dances with current AI technology. But the typical worker bee has about a million neurons in its little brain (and consumes only a tiny amount of power doing so). LLMs with trillions of parameters should be able to model this behavior if we were on the right track and seeing real emergent behavior.

The people arguing that they are seeing emergent behavior are anthropomorphizing, which humans are really good at doing when they want to.

At the moment its just an epic search engine that humanises the results, but
No, most AIs are trained on subsets of the internet at a particular time. If you want more current stuff you have to retrain the whole network at enormous cost.

A better way to think of LLMs right now is absurdly complex auto-correct and auto-complete.

You cannot have an interaction like this on any topic with any current LLM:

YOU: Can you show me how to do <this> and <that>
AI: <gives wrong and weird answer>
YOU: That's wrong. You need to do it differently. <gives an example>
AI: <gives a still-wrong but somewhat better answer>
YOU: That's a little better. Here are some more examples: <give more examples>
AI: <finally gives a halfway decent answer>

Current AIs cannot learn outside of their very expensive training phase. I'll argue to death that you won't have a general AI until it can continuously learn. Which is still something that theoreticians argue about and some dudes play around with in the lab, but there aren't any realistic solutions yet.

The actual king of AI is named after the glass seeing balls from LOTR.
Palantir is more of a data science and analysis company that is rebranding itself as AI to take advantage of the hype. "Data science" is AIs geeky and awkward older sister. Palantir is also the prototypical example of creepy AI. I see a whole universe of useful non-creepy applications that I'd like to see us exploit first.
 
Yes, Ive been following data AI, it is very impressive, mainly because it just gives you the scientific facts.
How many, how long, how hot , how cold..not people say etc.
All Teslas talk to each other to learn the road for self driving and it took them from absolute radar positioning to 'understanding' simple vision, with Tesla ditching Lidar.
I always said a true self driving car would have to be conscious, there are too many things only a human would understand.
But holy smokes, the latest machine learning that can run in a car computer is approaching true human context.
The wave of construction workers, road rage, strange things in the road, its starting to cope and Musk will put that in his robots, in ten years time people will cry when a robot dies.
Thats near enough for me, the human brain can be mimicked, it turnsvout its not as hard as we thought.
Waver..
Im distancing my Musk supported tech advances from his politics, which is off the scale controversial
 
Last edited:
Back