Heads up!!! Here’s your massive holiday sports tech deals list! This includes the Garmin Epix for just $419, the Apple Watch Ultra 2 for just $699, the GoPro Hero 12 Black for $249, (or the brand new Hero 13 or $299!) and plenty more! Go check out the full list of 100+ deals here!
I’m DC RAINMAKER…
Here’s how to save!
Wanna save some cash and support the site? These companies help support the site! With Backcountry.com or Competitive Cyclist with either the coupon code DCRAINMAKER for first time users saving 15% on applicable products.
You can also pick-up tons of gear at REI via these links, which is a long-time supporter as well:
Alternatively, for everything else on the planet, simply buy your goods from Amazon via the link below and I get a tiny bit back as an Amazon Associate. No cost to you, easy as pie!
You can use the above link for any Amazon country and it (should) automatically redirect to your local Amazon site.
-
Check out my weekly podcast - with DesFit, which is packed with both gadget and non-gadget goodness!
-
5 Easy Steps To The Site
1
In Depth Product Reviews
You probably stumbled upon here looking for a review of a sports gadget. If you’re trying to decide which unit to buy – check out my in-depth reviews section. Some reviews are over 60 pages long when printed out, with hundreds of photos! I aim to leave no stone unturned.
2
Read My Sports Gadget Recommendations.
Here’s my most recent GPS watch guide here, and cycling GPS computers here. Plus there are smart trainers here, all in these guides cover almost every category of sports gadgets out there. Looking for the equipment I use day-to-day? I also just put together my complete ‘Gear I Use’ equipment list, from swim to bike to run and everything in between (plus a few extra things). And to compliment that, here’s The Girl’s (my wife’s) list. Enjoy, and thanks for stopping by!
3
Have some fun in the travel section.
I travel a fair bit, both for work and for fun. Here’s a bunch of random trip reports and daily trip-logs that I’ve put together and posted. I’ve sorted it all by world geography, in an attempt to make it easy to figure out where I’ve been.
4
My Photography Gear: The Cameras/Drones/Action Cams I Use Daily
The most common question I receive outside of the “what’s the best GPS watch for me” variant, are photography-esq based. So in efforts to combat the amount of emails I need to sort through on a daily basis, I’ve complied this “My Photography Gear” post for your curious minds (including drones & action cams!)! It’s a nice break from the day-to-day sports-tech talk, and I hope you get something out of it!
5
The Swim/Bike/Run Gear I Use List
Many readers stumble into my website in search of information on the latest and greatest sports tech products. But at the end of the day, you might just be wondering “What does Ray use when not testing new products?”. So here is the most up to date list of products I like and fit the bill for me and my training needs best! DC Rainmaker 2023 swim, bike, run, and general gear list. But wait, are you a female and feel like these things might not apply to you? If that’s the case (but certainly not saying my choices aren’t good for women), and you just want to see a different gear junkies “picks”, check out The Girl’s Gear Guide too.
I’d love to know what happened to route collections – promised at Camp Strava in May 2023.
I know this would be difficult and in the grand scheme, I personally do not care because I view my cycling as a “me” thing/ the ONLY thing that matters to me is if it is a PR or not. But most nonclimbing segments are probably owned by group rides. There is no way for individuals to overcome that group aero effect. There should be some way to identify groups to really make segments meaningful. Again, whatever, just an idle thought
No comments on Suunto’s ZoneSense announcement? Instead talking Strava? Tough times to be alive
I haven’t quite figured out your deal. Changing your user name to (try and) make it appear as different people asking for it?
Ultimately, only a single other (real) person asked about it. Just one.
It’s on my to-do list to look at, but honestly isn’t super high up a long list of high-priority things people are actually asking about (and yes, lots of people are asking about Strava’s issues). It’s solidly a rainy day project, especially because aside from just re-posting a press release, it requires a lot of work to figure out if it’s actually useful, or just another fad.
I’m with you, Ray. And, honestly, I don’t think Nick bester (aka Brad?) should be poking too hard for your review because he may not like the results.
Like a lot of Suunto users, I’ve been trying to figure out ZoneSense since its release and can’t quite grasp how and when to use it. There seem to be a lot of parameters that must be met for it to function as intended. And it doesn’t help that Suunto’s introduction to the feature was a dry, one hour(!) pseudo-scientific video.
I was rooting for ZoneSense when it was announced, but now I am of the opinion that Suunto was sold some algorithms by a health/tech startup and now feels forced to use them even if the average user can’t make heads or tails of it. Hopefully it’ll be useful somewhere down the line.
Anywho, I guess I am actually with Nick/Brad in that I would like to see your take on it.
(To be clear, I don’t think this person is Brad. Brad is a good egg.)
@BKensington
Genuine query: I’m not sure why Suunto ZoneSense is hard to understand.
With a recent Suunto watch, an RR chest strap, and one decent calibration workout, Aet and AnT are calculated dynamically for each sport. IMHO, The information is presented in an easy-to-understand way on the watch and smartphone apps.
The maths behind it is surely complex.
Whether or not the info is correct is debatable. The studies cited are small.
I’ve used it a fair bit, and I can’t say it’s wrong. Hopefully, @InigoTolosa will modify his dfa a1 CIQ tool with something similar for Garmin
Oops! Sorry, Brad! I obviously don’t know Brad (or Nick).
@tfk the5krunner – I understand how ZoneSense is supposed to work, but don’t agree that the info is well presented. In most of the graphs you’ll see posted by users, and Suunto’s own examples, the ZoneSense data seems to jump up and down quite erratically. Am I supposed to believe my body jumps in and out of aerobic/anaerobic states in the same way? I don’t think it does. But, maybe?
For example, nearly every time I start into a steep hill climb on a long run, ZS takes a huge nosedive, suggesting my body is entering an even lower aerobic state despite a significant increase in heart rate and effort. (You’ll see this on nearly every example posted by users: HR up, ZS down). I can sort of understand that this might represent an initial increase in efficiency as my body adapts to the change, but it takes far too long for ZS to catch up to reality. By the time I’ve completed the climb, and my HR has spiked to its zone 4/5 threshold, ZS may have caught up and register the effort as anaerobic, but by then it’s too late. I’m past the point where that info would have been useful.
So if I’m using the ZS app on the watch, when do I look at it to get a proper reading? Is there a specific time delay? Three minutes into a change in effort? Five minutes? Or is it supposed to be instant? A graph that bounces up and down suggests that it’s trying to give instant readings, but Suunto doesn’t say.
The other issue is the requirements to get accurate readings. I don’t think ZS requires only “one decent workout” to calibrate, as you suggested. As I understand, the first 10 minutes of EVERY workout are used for calibration. This means you need to perform a proper 10 minute warmup at the start of each activity you plan to use ZS, otherwise the data is skewed and useless. How do I know what a proper warm up is? Ten minutes in zone 2? Which zone 2? My pre-defined HR zone 2? Or the zone 2 calculated by ZS during my previous activity? Again, Suunto doesn’t say.
I love the idea behind ZoneSense. It’d be great not to have to fuss with HR zones and have my thresholds automatically calibrated every time I go out for a run. Personally, I don’t yet have faith in what I’m seeing from ZS right now. And I don’t think Suunto has done enough to make it user friendly.
hi
Just my 2c on this:
Yes, there are 10 minutes of calibration per workout. I wasn’t aware there was a requirement to do any kind of formal warmup. (The pre-existing dfa a1 tools were similar, needing X minutes of data to get the calculations working.) The first workout per sport will appear calibrated, but from my experience, it wasn’t. The second workout seemed to work as it should.
IIRC, Suunto expressly excluded intervals/hill reps, so it’s for steady-state(ish) efforts. For example, if my calf makes it to the start line, I’ll use it for an HM this weekend. I expect it to have some use, perhaps similar to muscle oxygen data.
Well presented: I’ll clarify that the watch app is well presented and actionable, IMO. I don’t have an opinion on the chart.
time delay: I don’t think there’s a time delay as such in the calculation (I’m sure there is one to some degree), but there will be delays in what’s happening with your physiology and with the calculation/transmission
Yes, Suunto did make it clear that ZS is not intended for interval training. Perhaps I’m just not a good fit for it, though, as all of my runs include some sort of elevation change and, as a result, frequent HR spikes.
I think my main problem (with the on-watch app specifically) is the frequency with which the ZS reading changes. I’ll look at it one second and it will show low aerobic, and, then 10-20 seconds later, it says high anaerobic even though my RPE didn’t change. Perhaps the algos just need to be smoothed a bit? It does appear ZS works better for cycling and swimming, activities with less severe spikes and changes in HR than trail running.
I’m usually a hard sell on new things, so maybe I’m expecting too much. It is early days, though, so I’m still rooting for Suunto and ZS.
I looked a bit at the 1 hour video and one thing they mentioned is that one shouldn’t judge the effort based on the graph itself, what matters is the 3 zones. So graphically one could be at the top of aerobic zone, but that doesn’t mean you should ease up to stay in that zone. So my immediate reaction is that they chose the wrong UX for the job. Don’t show me a graph if it’s meaningless for my user experience. I don’t need that to understand that it’s working under the hood, just give me 3 coloured squares with the active one pulsing, or a semaphore system, something like that.
Anyway, I’m also interested in the system if it can establish the right zones for me that specific day (stress, coffee, fatigue, heat will change them day to day). So far I haven’t seen real life reports from users and I’m afraid this might just be a case of moving the goal posts around, till it appears the algorithm is hitting something.
Here Brad 😜
I don’t really understand why people criticize other people when something works or they like it… I always publish humbly and without the intention of bothering anyone.
I rescued this post from a Suunto community and it seems to be amazing Zonesense vs lactate:
🔺Suunto ZoneSense VS Lactate test
Scheduled training: 5K War + 6×2000 3:55 + 1’5k cool down.
Conditions: storm, wind and rain. I had to stop in the first series of 2000 due to the storm, half an hour.
Nutritional conditions: No carbohydrate loading the day before. Hungry. Without eating anything during and only drinking water at the end.
Heart rate: I had about 4-5 heart rates higher than normal at these paces. Maybe the time or the lack of carbohydrates.
INTERVAL DATA:
1st 2000: 3:54 / 154BPM. 3.7Lactate
2nd 2000: 3:54 / 150BPM POST STOP
3rd 2000: 3:52 / 157BPM. 4.0 Lactate
4° 2000: 3:53 / 161PPM
5° 2000: 3:53 / 161PPM 4.1 Lactate
6° 2000: 3:55 / 161PPM 3.8 Lactate
Post cool down 1’5kms : 1.6 Lactate
CONCLUSION:
Training according to lactate at the edge of threshold
Zonesense : Reflects in each series that I have been between LT1 and LT2, without going over LT2 for a few seconds.
The yellow zone, reflects that the intervals have been high aerobic at the limit of LT2, according to lactate data and the theory of 4 millimoles to enter LT2
If we compare zonesense with lactate, this corroborates this.
Luis Carrión: HR has nothing to do with real effort, clearly proven today.
⌚️ Suunto Race
🏃 @carriongluis
As you can see it is working, the value are in the users who are using it.
Lol it looks like the Suunto brad is being famous everywhere.
Personally, ZoneSense is a Gimmick, you have to train like it’s way, if not ZoneSense cannot interpret it correctly.
But I want you to comment a little bit on Suunto new privacy policy, it’s a new bomb.
We have updated our Privacy Policy, to reflect the upcoming organizational changes over the next three months, after which the Group companies of Suunto in China will provide customer services, repair services, web operations, mobile & cloud development, and other key functions, and jointly collect and process your data with Suunto Oy
Having spoken with Suunto, they confirm:
1. each sport needs an easy aerobic workout for a one-off baseline calibration
2. each workout needs 10 minutes of daily calibration. This calibration must be for the same sport (i.e., you can’t do a cycling workout as a warmup for a run using run mode). This needs to be something approximating an easy aerobic effort ie you can’t simply walk around the house for 10 minutes in run mode before going on a run, a 10-minute jog would be fine.
The attached chart is a threshold effort from me. I didn’t warm up immediately before the run, which might explain the blip at the start. the run FELT like a broadly consistent effort for me. heart rate (black) shows it doesn’t track effort, Power (grey) is a closer match.
Suunto accepts that there are aspects of ddfa you need to be aware of to find it useful. they give several scenarios and athlete ability levels that they believe will benefit.
The lack over club features and controls right now bothers me most. Simply not being able to have a club, with specific routes available or engagement functions.
I don’t mind the quick edit feature when opening the app after completing an activity, but I wish it would allow me to choose what fields I want in it. The main thing I edit when I finish a regular run is selecting which shoes I used so that I can accurately track mileage. As it is, I have to go into advanced edit every time to get this selection.
I don’t think I’ve ever seen the quick edit feature, and I’ve been on the beta track for the app for a while. Maybe it’s because ActivityFix gets there first?
One cute thing I noticed this morning is with the new Relive-like playback feature. Today because of weather I did a ride in Zwift, on Watopia. Because the GPS track is set in the Solomon Islands, the replay showed my track on an actual island, which has nothing much to do with Watopia. Maybe Strava should use Zwift maps for this feature for a Zwift ride.
The quick edit lacks one main function: changing the sport from the one recorded in the activity, which many have to do with older units that don’t have a wide range of sports built-in (Garmin lacking hiking and recording as walking being my prime and frequent example).
So it becomes a not-so-quick edit feature.
I have some open tickets on support.strava.com for them to mass flag or suspend users who continually upload dozen/hundreds of activities in vehicles, even after being alerted by the community in the comments. Currently sitting at 1mth+ no response at all. Submitted many of these in the past and they would usually action in a week or so…
Strava has always had problems with drawing a route. I’ve always used ridewithgps.com, it’s easy to build a route. Easy preview to google street.
Regarding Athlete Intelligence and that Strava should maybe look into generated title suggestions: Groupetto does exactly that. We see this as a convenience feature and it just makes sense. After all, many people don’t want to think a lot about titles and typing them in. Yes, we are still improving this feature, but take a look at where we are and if you have feedback, get back to us! (Still hoping for a review, Ray.)
link to apps.apple.com
Why can’t you still view the full list of players for a segment on mobile Strava? It’s boring that I have to go to the website and view the entire tables there.
Let Strava focus on serious priorities.
Great Fit File and I found it entertaining. I could have watched you both roast Strava for twice as long!
My thoughts:
Athletic Intelligence
I have serious concerns about how Strava has implemented Athletic Intelligence, it just seems to be looking at stats in isolation and not as a whole package. Take the average speed for example, if I told my 9 year old daughter that my last ride had an average speed 3kph higher than my 30 day average she would tell me that’s great. But if I told her I was on the flat and didn’t go up any hills she would tell me that’s cheating “Even I know going up hills is harder daddy.”
Strava has billions of data points that they can use to train an AI/ML model on and it seems to me they haven’t actually built any model at all. They pull a stat from your latest ride (be it HR, Candace, Avg Speed), run a query against the previous 30 day average and bingo they throw out a generic text. It seems to me they are more interested in getting an AI feature on the app more than developing a real and useful AI implementation. From an AI perspective I would expect some of these feature:
I did a hard ride yesterday so when I did today’s recovery ride and uploaded it then the AI would give me feedback on how well I did on my recovery ride. Did I push too hard? Did I take it too easy?
I did an interval ride today and uploaded it. Strava would recognize that from the warm up, 5x intervals then cool down. It would go away and find other interval rides I did and give me useful insights based on all the available data. Some rides could be on the flat some on rolling terrain, but formulas exist to normalize/equalize them…e.g wattscalculator.com
To be honest, I could jump on a plane to France hire a basic road bike and descend any of the iconic climbs or descents seen in the Tour de France and I’m sure that my average speed will be higher than any ride I have completed in the past 30 days. Strava’s Athletic Intelligence would then tell me that I’ve nailed it and my training and hard work has paid off. It wouldn’t look at things like cadence, heart rate and power (not that I have a power meter) or relative effort. It would totally ignore the profile of the ride.
One other concern about the new Athletic Intelligence is that it currently only runs on the mobile app, not the web. I’m pretty sure a lot of the smallest AI LLM models are still too big to run on my mobile phone which tells me that its not a true AI implementation.
Anything that Athletic Intelligence currently gives me I can get from Garmin Connect on the web by running reports. I can see trends, spot outliers, I understand my own body and realize my heart rate is trending down because I’m getting fitter, I’m going slower this week because I’ve climbed more or my heart rate is higher because I was at the pub last night for my brothers birthday and hungover.
Strava’s AI implementation is a massive fail!
Quick Edit
Leaving behind the AI, how did the new edit on the mobile app pass any basic QA checks? And how did it even get designed in the first place. Strava is a software technology company that deals in data. Any update to software needs to be follow certain processes to be successful. I am at a loss to see what the business requirements were? Did they document any use cases? What were the developers given to go away and write the code? Do they have a test team to verify the use cases? have they outsourced the development like they have the PR team, community and forums? it certainly seems that way.
The UX is awful to the point of being annoying. I certainly would never let any of my teams release that to internal QA teams never mind public beta testers. Why force such an intrusive experience on app users in the first place? A better UX would be the following:
Given an activity has been uploaded to Strava for a user
When the user views that activity on the mobile app for the first time
Then a pop up is displayed asking if the user wishes to edit it
Given multiple activities have been uploaded to Strava for a user
And a number seem to be similar but with different devices
When the user views one of the similar activity on the mobile app for the first time
And they edit the activity
Then prompt the user if they wish to apply the settings to all similar activities
Given a few hours dedicated to thinking about the UX I could write dozens of scenarios that would make more sense and be more user friendly than what they have implemented.
I’m with Des and Ray, give me a toggle in my profile so I can switch it off completely. Even better would be to roll back the change until they have actually done some analysis regarding what UX they wish to implement and test it internally before just throwing it out in the wild and hoping it works.
KOM clean up
Another massive fail! Given that they have millions of data points I am sure that any decent AI/ML developers culd design and implement a model that would highlight fraudulent KOM/QOM entries. A few climbs that I complete on are similar in ascent and length to the spring classics, yet I am sure even Matthieu van der Poel could not claim the KOM. Strava is planning to implement verified segments, what about verified riders? If a top 10 classics rider can complete a 1.5km 10% average climbs in x seconds with y watts in a classic then “Big George” completing a 99% similar climb of 1.5km 10% average over a 45 seconds quicker on his local route must be flagged. I’m sure Big George could attack Tadej Pogacar and win the world champs.
I was shocked the other day on one segment, I had the 6h best time this year and 69th over all, must be a mistake! Second in my age group. Not bad considering I’m in my 50’s and overweight. Admittedly I had a tailwind most of the way at 27kph which helped. Yet another segment later in my ride I was nowhere near that level, similar elevation and direction but double the length, and my average speed was almost identical. I was miles off any leader board! KOM 24kph on average faster and average HR less than 90bpm, second place is almost 150bpm. Someone else on the top 10 has over 700 watts over almost a 3 minute effort, everyone else round about has half that. WTF?!? No matter what anyone says stats do not lie. Politicians bullshit the story to fit the stats, but the stats dont lie if you have intelligence and examine them!
Strava has the data, they know whats normal, whats super human (Tadej, Matthieu, Remco), and whats downright bullshit and cheating!
Fatmap
Sorry cannot comment and I didn’t even know about it, but seems on the face of it as another fail.
Strava Routes
Yip i agree they are messed up. i tend to explore more using heat maps then save an activity as a route if I like it, but I find Komoot is my go to routing platform these days.
Finally
This seems like a big WTH moment for Strava. Have they outsourced the design and development teams to a team who does not know what Strava means to the community? Have they been smoking something? Who knows! Hopefully they can get back on track.
Whenever I race, I put a link in the Strava activity to the external race results website. These links constitute a really useful record for record for me. For Strava to just go an remove all these links from my record is deeply upsetting. I will be hard pressed to find all these results again! It is especially bothersome given their terms and conditions say that this record is mine! Of course when I emailed them to ask why they couldn’t have just made the activities private or shifted the links into private notes (where they are still allowed), I got no response. This is enough to make me reluctant to renew my subscription,
I first noticed in the profile description a few days ago that the links had been removed. Thanks to the article here and your comment, I realized that Strava had deleted all the links from all the old activities. Not even a placeholder or reference to a deleted link was inserted.
Was this communicated by Strava? That is incredibly cheeky of Strava in my opinion. Luckily, I still save all my tour reports in my own database.
But I can perhaps give you a tip. I haven’t checked it yet. It could be that Strava hasn’t deleted the links in its database at all, but just doesn’t display them. Download your entire profile as an archive. It could be that all the links in the activity descriptions are still there.
With the new quick update. Is the activity posted before I complete the quick edit??
Watched the FitFile on Strava route builder issues a few days ago. I use Route Builder often. Found it easier to use than Garmin’s builder a long time ago and have always used Strava and let it sync to my Garmin courses.
That SHOULD be a good thing for Strava, to have more courses drawn out for them (since I always make them public, anyone could use them.)
Wondered how it would go next time I used it… Last night I needed to create a course quickly to send to a riding friend for a century ride we did today. Should have been simple, and always has been before.
NO MORE!
The route builder I’m using is on a laptop PC. I found the phone app not very friendly by comparison and and have used it minimally.
TOTALLY UNUSABLE ROUTE BUILDER!
Started a RIDE, with Prefer Paved Surfaces and Popularity routing.
I placed my starting point at one end of Douthat Rd, (SR629), a paved road that runs north for 20 miles to connect to Rt39. It is entirely paved and has been for all the 40 years we’ve lived in this area of Virginia, USA. I placed my next point at the north end of the road at Rt 39.
Instead of following Douthat Rd straight to the next point, STRAVA went OPPOSITE direction (south) and then west for 15+miles, then north for about 20+ miles on Rt220, then east up over a big climb on Rt39 and down a long descent to the other end of Douthat Rd. Instead of 20 miles, it was probably 50 or more.
I tried changing from Popularity routing to Direct. SAME RESULT.
Strava’s Map shows Douthat Rd as mostly paved but has a chunk of several miles showing as Unspecified road type.
I had MANY more roads to add to the century ride route I wanted to create and share with my fellow rider, but after wasting time trying to get it to work, and the only solution was going to be adding via points in the middle of long stretches of road to try to FORCE Strava to follow the OBVIOUS choice (which IS a popular route, in fact an excellent choice for low traffic, decent surface, moderate climbing, etc), I FINALLY gave up completely on Strava and created the route in Garmin and shared it with a link. (We rode it today and it was a GREAT, tough century on a gorgeous autumn day!)
So I’m BAFFLED at what Strava was thinking with this, or what is causing the nightmares, but it absolutely won’t keep me as a subscriber if it continues to operate so poorly. I use this feature as much as any on Strava, and need it to work dependably and consistently, and it has done so for quite some time, at least on the PC app.
And get this, I JUST NOW TRIED IT AGAIN because I wanted to see the exact term Strava uses for unknown surfaces (UNSPECIFIED), so I duplicated what I tried last night, and GUESS WHAT! It WORKED EXACTLY AS IT USED TO, and AS IT SHOULD! No goofy behavior at all. ???
Is it FIXED NOW???
Time will tell!
I created a couple of short hikes last night on Strava and had only the usual amount of difficulty, coaxing onto the trails I wanted rather than weird out and backs, etc. I got the loops I wanted eventually. So maybe they did fix it. The last ride I created was a couple of weeks ago, and there again only the usual amount of difficulty. (Never tell Strava that a ride is gravel or mountain, because then it shows an inordinate fondness for single track and will go out of its way to include gravel and/or trails) I hope I missed the problems entirely, and they don’t come back, because Strava is my usual go to for routing.
Here’s a weird thing about the routing…
Mapping a trail run near me Strava does everything to avoid a 100 metre stretch of path (as in the FIT File), BUT maps it fine if I reset routing preferences to Bike (despite these being v rough and narrow paths).
Difference is massive. Swap it back to Trail Run and it triples both the length and the elevation, and takes you on a crazy series of diversions. Run, Walk, and Hike all plot a third (slightly less) crazy route.
Makes no sense.
I only hope they will not raise the subscription price (again) just to pay the bills of the energy wasted on the AI bandwagon… this would be unacceptable for me.
Actually Strava is getting less and less interesting now.
One more thing that they’re not able to do is search for activities: they need a full word, in the exact same collation used to write it. So any ümlaut in a word can make the activity impossible to find. And again: only full word search…
By the way, about the recent enshitification of Strava’s routing tool, and further to the comment I left along your FIT File video on YouTube, I did a bit of digging around.
Turns out, I think Strava used some sort of AI processing to “improve” their terrain data: around where I live in France, there are a load of paved roads called “Chemin de [whatever]}”.
Chemin meaning originally “track” or “path” or “trail”, often used also for farm roads etc.
Before the recent Mapocalypse, those were correctly routed as paved roads (and they are still displayed as roads on the Strava map itself), but now if you route through one of them, the trace appears as the red/white dotted line that means “trail”. So it’s classified as “not paved road”. And so, roads that used to be OK when designing a “road cycling” route are now not proposed anymore, creating those crazy routing detours.
Seems to me they got some machine learning/genAI tool to look at road names and try to deduct the terrain (so “chemin” means “path/track/trail” means unpaved, rather than just relying on the map dataset’s terrain layer (presumably OSM?), that used to be very good.