Why car wheels are so flat these days
theautopian.com
theautopian.com
Here's their writeup on the electric F-150 - keep scrolling on that page, and look at the kind of depth and detail they get into: https://www.theautopian.com/the-2022-ford-lightning-is-just-...
The site is a gem.
Secondly, it is perfectly possible to design cars that do not need tons of invasive repairs from minor fender benders. Revise bumper and front/rear cosmetic design to allow for effective bumpers (see also: the 1980s) and tune crumple zones to need higher forces to deform them so you don't have to potentially replace a whole car over a 10-15mph collision. The tradeoff here is that the speed range at which crumple zones do much would move up which is good for real world performance bad for scoring that perfect five stars in a low speed lab test that actually helps sell cars.
Can you please cite where I stated, or hell, I'll settle for strongly implied, "the safety tests are invalid" or that "the engineers designing cars don't know what they're doing".
The person I am replying to made a trite low effort comment. I, without calling them out like they deserved, explained that the situation is more nuanced and their opinion is slightly off mark. You then replied with another trite low effort comment straw-manning me. FFS this is ridiculous.
You also said that designers should, "Revise bumper and front/rear cosmetic design to allow for effective bumpers (see also: the 1980s) and tune crumple zones to need higher forces to deform them so you don't have to potentially replace a whole car over a 10-15mph collision." This strongly implies that the bumpers as they are currently designed are crumpling needlessly at low speeds, and that they could be made to not do so, like the bumpers of the 1980s, without compromising safety, hence that the engineers designing them are not competent. Although admittedly in retrospect it could also be that the engineers are designing to the aforementioned tests. Regardless it's an extraordinary claim, requiring evidence.
Your first paragraph also makes a number of claims, many of which I'm sure are true. Certainly curtain air bags and seat belt pre-tensioners have been shown to save lives. I recall reading a study on the positive impacts of adding pre-tensioners to the back-seats of vehicles a while back. But again I don't see anything supporting the claim that crumple zones are ineffective, or specifically, that, "The speed window at which they will appreciably reduce the deceleration in the cabin is well below the speeds at which a seat-belted occupant of an airbag equipped vehicle has to worry about serious injury." That could be true, but it's not supported here.
https://www.youtube.com/playlist?list=PLGSOZAHg1yQHU1tc_3Y5M...
Thought this was interesting: https://www.theautopian.com/this-junky-postal-jeep-with-a-ch...
Looking at photos of the F-150 Lightning chassis, it seems like a pickup truck's body-on-frame design has a nice unintended benefit of being easy to convert to electric. The "frame" doesn't look so different from the "skateboard" EV design that's common.
A lot of converted gas to EV cars like the Genesis Electrified GV80 rob interior space for shoehorned batteries, but the F-150 is all gains with that massive front trunk.
Also, the color on those wheels makes me think of a 90s Impreza (paint job in blue, please!).
Braking redundancy will be achieved by having motors/brakes on all four wheels, and within each motor 3 independant phase coils with independant controllers, such that there are effectively 12 brakes on a car. Normally the controllers work together for smooth braking, traction control, software differential, etc. But even after 3+ failures braking performance should still be satisfactory for an emergency stop.
Obviously braking energy needs to go somewhere. In the happy case, it's regen'ed into a battery. If the battery can't accept it, it gets dumped into dump resistors. If the dump resistors fail, it gets dumped into motor coils (of which there are 12 remember). Obviously the motor coils will heat up very fast, so this is probably a one-use-only failsafe, like airbags.
So the whole system (except the pedal itself) is 12 way redundant.
iirc both times it was some sort of brake fluid leak.
So, while I totally agree that having motorized breaks is really scary, hydraulic brakes can fail as well.
The cable-operated E-brake definitely doesn't stop the car anywhere near as well as the power-assisted hydraulic brakes, but having two different braking methods can be a literal lifesaver when one system fails. (And all systems fail! I'd previously had two other cars where the E-brake didn't work.)
Good on you.
You are probably referring to the brake assist mechanism in all cars after the 90's. It uses vacuum pressure from the engine to help you push the pedal down. Without the engine running, the brakes will work ~3 times, and then they'll get super hard to press - but they will still function if you have leg muscles strong enough to work them.
Just a friendly reminder that your car has brakes. When they break, then they will fail.
Planes have a whole different set of expectations on their reliability and upkeep.
For example I don’t think people would be excited about an electric car with an aux power unit.
All of that can be done with software-only - no hardware changes needed.
The slam on brakes case would only trigger if there was gross damage to the braking system, where you might have bigger things to worry about.
Control power is different from actuation power, the latter of which is what you're addressing. And it's not "gross damage" because it's a common-mode failure, so you can't just ignore it like that.
Edit: made more concise.
This is still uncontrollable failure. What if the car is in the third lane doing enough speed and there's some cars behind? I assume no ABS, because ABS can't work with locked up brakes. This means no steering, just skidding whatever way.
Having a stick shift definitely helped, though.
But there are ways you could walk away from this.
You see some pretty crazy crash survivals in race driving, for example.
And I remember stories from a few years ago, when Ford had a recall on F150 tires because they tended to blow out at highway speeds, with an effect similar to full wheel lockup-- yes, people died, but others didn't.
life == hope
Tesla's that lose battery are already a challenge to tow as the parking brake cannot be disengaged without power.
I would love to be present at the DFMEA (Design Failure Modes Effects Analysis) meeting when this came up!
If emergency braking could use an explosive-spike-in-ground deceleration at say 5 G, then far more crashes would be survivable.
I think the reason it hasn't been done are human reasons - users don't like the idea that jabbing the brake pedal hard will mean they have to pay tens of thousands of dollars to repave the road and replace their spike.
To some of the replies... I think some clarification needed because the wording caused folks to think of something else such as electronically _controlled_ physical brake pads: https://en.wikipedia.org/wiki/Electronic_parking_brake
I think the gp was trying to ask about feasibility of electric _powered_ magnetic field resistance braking ... similar to an exercise bike ... which would require constant electrical power source. (Similar to : https://www.google.com/search?q=magnetic+resistant+braking+o...)
Magnetic field resistance brakes is practical for regenerative braking but not safe for parking brakes on an incline if the batteries run out of juice. The context is the gp's premise of no physical friction brake components via "(ie. no hydraulics, drums, rotors, or pads)."
In my Tesla, I push a button and I can hear an electric motor engaging the brake.
The whole industry is working on tighter integration of the drive unit components to reduce things like cooling circuits, length of high voltage cable runs, and the total mass of the components required to move the car. No one is working on splitting the motors out into the hubs.
As you point out, unsprung mass is bad. Especially for something like a motor it's beyond bad, it's a complete killer of ride quality and reliability. As sprung mass the entire drive system can be isolated from all the knocks and oscillations that happen from driving around. A hub motor will get crashed and bashed around and have a miserable MTBF. The occupants will hate the car since it will have awful handling and hitting even moderate-sized bumps will be terribly unpleasant. Lighter stuff like scooters can get away with hub motors since there just isn't that much mass to begin with, but cars are huge now.
Aptera is planning to use hub motors, but their car is the vapourist of wares and relies on being counted as a motorcycle under regulations. Their basic engineering just doesn't work when scaled to a 3500 pound car and beyond.
https://lightyear.one/articles/lightyear-and-elaphe-develop-...
You should see what you propose on cheap cars if motor pricing becomes less than the cost of regular hydraulic brakes.
FVMSS 105: https://www.govinfo.gov/content/pkg/CFR-2011-title49-vol6/pd...
What is a "service brake"? Notice that the regulation you linked to includes requirements for EVs that have braking regeneration as a component of their "service brake".
It turns out that the top-level of FVMSS [1] says that "service brake means the primary mechanism designed to stop a motor vehicle."
So I think you are exactly right. There is nothing in the law that says you can't use a 100% brake regen system in an EV. It just has to work and meet the performance standards (and perform regen on all wheels). But I'd guess that you aren't going to see the traditional brake system disappear unless it starts getting in the way. They don't cost a lot (relative to the price of the car), don't weigh a lot, are really reliable, and work really well.
[1] https://www.ecfr.gov/current/title-49/subtitle-B/chapter-V/p...
I don't understand what you mean by this. Isn't independent suspension already a thing? Each wheel has independent up/down movement because there is no axle that remains straight (driveshafts are independent from each other in both FWD and RWD).
Tire friction usually works on a vector-like system so being able to move more up-down rather than along a radius means maybe you'd get a couple percent more friction forward and back if not moving a little side to side when moving up and down. Which sounds useless, but 5% more usable tire friction means an improvement in braking and cornering performance which either makes the car safer or could theoretically make it cheaper/lighter to compensate.
It's note even with resistors it's only with resistive braking that this applies.
If this is an electric car then the inverter can be programmed to give pretty much any braking force you desire just by altering the phase of the driving waveform. This can be done at any rotational speed.
My Model S can dump 50 kW back into the battery so long as the battery is not fully charged. At 60 km/h the kinetic energy of the car is about 300kJ so the car can be brought to a stop in about six seconds purely by regenerative braking.
In the final phase of braking the car has hardly any kinetic energy left so the huge 250 kW motors can easily bring the car to a halt and hold the car still just by balancing the magnetic forces in the motor against the motion of the car or gravity. In fact this is exactly what it does when running on traffic aware cruise control in stop go traffic, whether on level ground or on steep slopes.
I'm looking forward to the day where beamed power (e.g., MASER) will allow us to ditch most of the batteries too, with a tow trailer battery pack for trips into the deep country. It makes no sense to me to accelerate and decelerate over a ton of battery pack and have to deal with so much regeneration loss when we could just have overhead beamers optically track vehicles and cut power whenever a stray bird enters the picture.
I jest, of course, but I would kind of like to drive a car with steampunk levers more in theme with the giant spider in Wild Wild West.
I used to make Lego vehicles like that (well before I could drive), because it felt so silly as a passenger being parallel parked I suppose.
Replacing airbags (at least individually) is cheap enough. Having your entire engine-equivalent need a replacement just to reduce the sprung weight?
And maintenance would be way cheaper. Just remove the wheel and put a new one on.
* Regulation. Plenty of countries are very specific about the exact mechanicals of brakes, in laws that are 100+ years old from when runaway cars killed people.
* Unsprung mass. Cars handle better if the wheels are very light and all heavy bits are in the body of the car.
* Large diameter motors are harder to design - there is no opportunity for a gearbox, so you lose one design parameter, and need the motor magnetics to be right around the rim of the wheel to make up for it.
* Difficulty of making the air gap within the motor very thin at such large diameters - you end up needing super large diameter roller bearings.
However, I think despite those downsides, the benefits outweigh them.
1. Braking generates a LOT of energy which needs to be sent somewhere very quickly. Each wheel might need to dissipate hundreds of kW of power, simultaneously, on-demand, at any time in normal situations where mechanical brakes function fine today. This is not inexpensive.
2. Today, most EVs do not have 1 motor per wheel, which would be almost surely required to implement this type of system effectively. That's also not inexpensive and unless you can actually put the motor in the wheel-area you still need an axle to connect the motor to the wheel due to the suspension and steering movement, and then that axle would need to be able to sustain these braking forces.
It surely could be done from an engineering point of view today, it's just going to add tremendous cost over using conventional mechanical brakes regardless of if those brakes are a traditional hydraulic type or if they're a brake by wire system.
Not sure I'm convinced by the torque claim -- we've seen cars skidding / losing traction in both scenarios (accelerating and braking). ABS brakes were developed specifically to prevent that (and so was Traction Control).
My experience with electric vehicles is limited to a Boosted longboard, so take this with a grain of sand -- but it seems to me that electric motors are as effective at braking as they are at accelerating.
Putting a motor on each wheel could have some performance benefits, but I'm not sure I'd pay for it if I could get a single-motor car for cheaper. I'm talking about "econo box" performance, which is all I want anyway.
Also cars can stop way harder than any regen can, massive amounts of energy will be at play and likely any solution is more complex than just grabbing a rotor, which is a more direct way to heat dump
http://buick.oldcarmanualproject.com/manuals/1956/1956%20Bui...
Observe that the kingpin axis is entirely inboard of the tire, almost vertical, and the scrub radius is such that the axis meets the ground inboard of the contact patch of the tire (in other words, it doesn't really "scrub" the tire when turning).
What stops the suspension from actually steering is the steering system itself and the hands of the driver holding the wheel. You can visualize that in the picture above or below; if the ground pushes the tire rearwards at the contact patch, which is a few millimeters outboard of where the kingpin axis intersects with the ground, that tire will have a tendency to rotate, which will pull on the steering tie rod and send forces to the steering wheel.
What isn't mentioned is that if you're braking, then the wheel on the other side also has the same force applied to it; only the difference between the two forces is what attempts to back-drive the steering system.
Note that it is harder with rack-and-pinion to approximate https://en.wikipedia.org/wiki/Ackermann_steering_geometry which is advantageous for tire wear.
Personally, I prefer the smoother and more isolated steering feel of a traditional recirculating-ball system, but then again, I'm not the type of driver who likes fast cornering.
Until you got used to its incredibly sensitive response (about 1.5 turns lock-to-lock) you tended to drive in zig-zags like you were tacking a dinghy up the road, but once you did get used to it nothing ever felt as precise or comfortable. I've driven some incredibly "sporty" cars, "cost as much as a house" end of the market cars, and none of them got close to the CX for steering feel.
The earlier GS and GSA had inboard front disc brakes on the side of the gearbox, allowing them to be massive while also allowing the tall skinny tyres (145SR80, about 150mm wide) to be absolutely centred on the kingpin, boresighted on the middle of the contact patch. They had incredible handling, so sharp and responsive, and the non-powered steering was as light as any modern car at parking speeds.
It does tend to be more apparent in front wheel drive cars, perhaps it's the weight of the motor? It's really important to get right in motorsport, as it denotes the characteristics of the car when you've oversteered. A car with correctly set caster will "self steer" itself back to straight without any driver input, and it will guide the car while oversteered in a way that's balanced with the direction the car's actually heading rather than where it's pointing (if you're intending to be sideways, like in drifting).
You read the title and think "D'oh obviously traction", then you see the first picture in the article and think "Yep, obviously flatter tyres because traction", then you read a little more and think "Wait, what do they mean by flat?", ...
... and then it just keeps going deeper and deeper into this rabbit hole you never considered before and barely even noticed. You realize you didn't know shit.
The explanation ends up being steering feedback forces. Now I wonder if we can have deeper wheels again with modern electric motor power steering.
[^1] https://thedecisionlab.com/biases/the-illusion-of-explanator...
.... which, is used to communicate traction status back to the driver:
> The ability of forces to feed back to the driver meant there was significantly more “feel” for what the tires were doing and where the limits of adhesion were. (from the article)
It can also break the driver's thumb, if the thumb is on the inside of the rim. If you're driving over rough terrain, you're well advised to keep your thumbs on the outside of the rim. This is true even at low speed. I once had to pull a car up onto a normal-sized curb and did so at low speed. Doing so definitely kicked the wheel.
I do very much enjoy watching “Kid Stance” videos on YouTube if you want to see a really brilliant creative guy make some AMAZING custom kid ride on toys…warning: addictive!
It's not the tires that are flat, it's the face of the rims.
The parts of the suspension the wheel turns on are pushed out towards the center of the contact patch on newer cars to reduce the forces that feed back to the steering wheel. The different steering system used on older cars benefited from those larger forces (because the systems didn't have good feedback).
They're talking about the wheels, not the tyres.
I've always driven cars with rack and pinion steering while wheels went from something like a typical 185-65x15 to 245-40x18. So, wheels are larger and the profile is much smaller (more rim and less rubber between the car and the road).
Higher power warrants bigger brakes and bigger brakes warrants larger rims but for the average car rims have grown way oversized. Many cars have the space to take rims that are 2-3 inches smaller than what comes as standard.
My anecdotal and average empirical results is that cars with wide rims and little rubber mostly seem to reduce comfort, increase noise, and make the steering very susceptible to unevenness, slopes and grooves of roads and layers of snow and ice and thus requiring constant steering response from the driver. I've had both kinds and the difference is huge, and exaggerated by bad road conditions that we find on our roads. On my older vehicles with a higher and narrower tyre profile I barely have to steer at all but just keep my hand relaxed on the wheel while the car ploughs pretty much through whatever in a direct line. Of course, I keep the wheel geometry of all my cars well-aligned so they shouldn't wander around.
> just keep my hand relaxed on the wheel while the car ploughs pretty much through whatever in a direct line
I would avoid small potholes still but that's basically my feeling on that car. I did slightly bend a rim once when running through a pothole I did not see, and the car ploughed through as described.
I recall reading an article a couple of months ago about steering, feedback, and suspension geometry design. It rang very true to my subjective experience that many car suspensions are designed for "comfort" (vs "sporty"), which results in a lack of relevant direct feedback, which makes the car uneven and floaty, like you're somehow disconnected from the road, and still suffer from bumpiness as you have to rely on indirect cues to compensate, which makes for increased latency and overcorrection.
I wish I could find this article again, IIRC it all comes down to one angle/offset, which you can't really fix by tuning geometry because it's basically baked in the suspension design.
EDIT: it's related to that one[0], but was more detailed regarding that part:
> To make matters worse, the suspension geometry often further compromises steering feel in favor of things like putting less load into tie-rods
[0]: https://arstechnica.com/cars/2022/01/new-cars-have-lost-thei...
Additionally, if you move the wheel surface far enough from the center of the tire you can get more room for wheel bearing and axle shaft hardware without increasing the size of the wheel bolt circle and bore though this isn't a big consideration in a lot of applications as there is ample space either way.
if you want to say, "you can't use a cone like that because there needs to be a hinge on the axle at this point for the wheel to steer-turn" then we've arrived at the reason for the limit on deep dishes? or if it's some other reason, it still doesn't seem that hard to describe.
what I'm really saying is, the geometric concepts could be taught more simply without the all terminology. My older brother raced cars that he fixed and modified himself, so I grew up basically in a mechanic's shop. That was a long time ago so it's fine if terminology and concepts have changed, but I have a big head start over the average person and I found the article annoyingly incomprehensible
The steering is definitely different, especially because I also have tires that are 3 inches wider than the factory ones. The car is harder to keep straight if the ground is uneven because it wants to pull in all sorts of directions. I also don’t like how at full lock the outer wheel drags on the ground. But at this width the closest I can get is an offset of +35mm, any higher and the wheel touches the control arm, and at 40 it would probably hit the springs.
Talking to some track guys seems like they are still faster with a wider track than with lower scrub and told me to not worry about it, but I still have my concerns.
Of course on undriven front wheels it wouldn't work because there is no axle, but in front-wheel-drive cars there is one. The brakes could be inboard of the CV joint, no?
https://www.dieselworldmag.com/wp-content/uploads/2017/03/DW...
This particular example doesn't even have front brakes.
1. Brakes generate massive amount of heat and need cooling. There is a ton of airflow down at the wheels, some wheels are even designed to direct more airflow to the brakes.
2. Because changing brakes down at the wheels takes 20 minutes with the proper tools. If you move them inboard and affix the rotor to the axle you would significantly increase labor time and cost in replacing the brakes.
Brakes are a wear item, they eat themselves by design and need to be replaceable. Moving them inboard from the wheel does not have any significant benefits, but many downsides.
With regenerative braking on EV's it could possibly make more sense to move the brakes inboard, but there is already all the space in the wheels for brakes to fit, so I don't really see the benefit of moving them.
There are a lot of downsides, like maintenance is a bit of a nightmare.
One I find particularly amusing is the Ford Mustang Mach-E, a fully battery-powered vehicle, has the ability to play fake engine noise through the cabin speakers because some drivers think engine noise is a feature of "real cars".
I’m pretty sure BMW was doing it too when they started going all turbo and losing the song of the NA.
But so isolated from the world around it that you can't hear it.
There a lot of things that make a car feel faster, and giving them higher displacement engines and louder exhausts has more disadvantages than it’s worth for most people.
You can get the same experience with slightly more cabin noise (that doesn’t annoy your neighbours), a longer first gear, kickdown pedal, etc.
It’s kinda like Coke Zero. If you can get 95% of the same experience, why opt for the sugar?
Thin wheels help with aquaplaning/hydroplaning (strongly) and slightly with reducing consumption.
Most wide wheels available to consumers are thick, so that is a tradeoff that happens with the new design, although I might mix up physical ratio, price and street legality, due to having only looked at one car.
The challenge is how much optimization for efficiency vs. durability vs. cost manufacturers want to or are ordered to sacrifice.
I tried to google for an explanation, but only found hits about brakes problems.
> [Editor’s Note: Not all designers prefer dished wheels. Some commenters were wondering why the Jeep JK/JL Wrangler, which has an old-school steering box instead of a rack and pinion setup, has such flat wheels. As far as I know, this is largely a styling decision. -DT].
So, it also applies the the XJ limited wheels too:
https://i.ibb.co/zfHMNbB/B03-FA738-777-A-40-D6-817-F-C0785-E...
There was this gem [1] posted in reddit the other day and I guess it is partly a failure of UI and inexperience.
A junior mechanic tried to fill the tires to "100%". https://i.redd.it/eu73w8earfn91.jpg
Edit: fixed, old reddit. App doesn't make old reddit links.
[1] https://old.reddit.com/r/Justrolledintotheshop/comments/xcdv...
Are the official consumption ratings not based on this kind of overinflation, removing all seats but the driver's?
So everyone is expected to do both these things by the car manufacturer, if that is 'realistic use'. /s
Now I got a Mazda CX5 and it has “just” 19” rims with tires that are wider than the rims.
Stuffed with filler words to the max. Maybe there is an answer to the title somewhere on the page. But the way it is written, it just stole my time until I gave up after a few paragraphs of filler content. First sentence:
Have you ever wondered why designers show
sketches of concept cars with massive deep dish wheels, but
when those cars actually make it to production the wheels
end up being fairly flat?
Compare that to a classic essay "A plan for spam" by Paul Graham:
http://paulgraham.com/spam.html
First sentence:
I think it's possible to stop spam,
and that content-based filters are the way to do it.
If it was 2022 style SEO content, PG's essay would have started with "Have you ever wondered why you get so much spam into your inbox these days and why that spam is so terribly boring, sprinkled with typos and ads, often full of images and links, and what you could do to get rid of it?". Shudder.
https://www.theautopian.com/this-is-why-car-designers-always...
Which was in part a reference to some questions on the comment threads from other articles in which a designer sketched some car ideas on that site.
So, you may not have wondered that, but someone on the site wondered that, and another writer on the site, who, granted, has an background in engineering car suspensions, not writing, wrote this article in response.
If you just want a straightforward answer then Google it and perhaps Google can find the right snippet that you can ingest in 8 seconds.
HN isn’t a feed for TekMol’s interests.
I'm not even a car person but this article wasn't laboring the point.
But don't get me started on them providing a rear-disc-brake Focus for the test drive even though the trim level I was interested in had rear drums. Had I driven the drum model, I would not have bought it. But I acknowledge that I'm an outlier, who actually cares about how the vehicle handles and accelerates and brakes and takes potholes and how noisy it is at highway speeds and how it fares in a cross wind. The common, typical "test drive" doesn't cover any of those aspects anymore.
I guess this is situated for latency but more likely installation and loom cost controls, close to the rear axle therefore more exposed to moving accident damage and at least to my casual reading of owner enthusiast fora unknown to educated owners and the public - rather like the Eurofighter Typhoon?
How can such designs pass DMV / MOT tests with inherent rolling instability in failure mode?
Or is there failsafe?
Edit: I should have said multi tonne vehicles intended by design to be driven fast and at limits including load limits.
Here is a picture of the part:
https://images.simplepart.com/images/parts/motor/parts/fulls...
I recently struggled a lot with a technical problem. I searched and searched and I couldn't find a solution. Then I discovered a solution myself and I created a short blog-style tutorial how to resolve it.
As far as I know the only one on the Web, but does it show in Google search results when one searches for exact phrase or quite rare keywords used? No way, I've added it to Google, Google scanned it, and that's it.
Searching for the problem you find my forum posts where I ask (unsuccessfully) for help about it and maybe few more people doing the same,nothing more.
Perhaps instead of writing my blog entry for humans (state the problem, describe a solution). I should've written it for Google bot in a format of:
- State a problem - why do we have this problem - what is the history of stuff we have the problem with - a personal story how I saved my dog while working on the solution - at the very end one sentence that hints at a solution (don't just give it away! Make 5 pages talking about it instead!)
That's why we can't have nice things.
I'm with you, I seriously hate that I have to read the internet that way nowadays.
That said, I understand Google's position. Back in the WildWildWest days of the WorldWideWeb, Google made the change from preferring keyword-based text to preferring written-for-humans text. But the target humans seem to those who read for love of reading, as judged by lit majors. Now I've been teased since kindergarten for my love of reading, but as a professional I want concise and terse, not an overabundance of anecdotes and adjectives. But the lit majors who influenced the post-2008 spider algorithms correctly for the time influenced a preference for the later.
Hopefully sometime Google will figure out that technical texts should be keyword based, and literary texts should be for-humans, whatever that is deemed to mean. And Google might provide either a feature to choose the algorithm, or more likely, try to guess based on heuristics.