Chinertown
Chinese Carbon Road Bikes => Road Bike Frames, Wheels & Components => Topic started by: Smokemachine on October 12, 2025, 11:32:52 AM
-
Bike Frame Index - My database on Chinese carbon frames
I discovered this forum in July and have spent an intense few months reading threads and bookmarking information while researching my own (road) frame purchase. To make sense of it all, I compiled my notes and bookmarks into a database that might be useful to others, especially people new to buying bike frames from China, perhaps doing their first road frame build: bikeframeindex (dot) com
What it is:
A summary of my research comparing popular Chinese carbon road frames on here, Weight Weenies, YouTube and Reddit. Each entry synthesizes what I found across hundreds of forum posts about geometry, quality, reliability, and value. It's of course quite biased toward my own priorities, and i'll underline that I am quite risk averse, but I've tried to capture trending issues and recurring themes from community discussions.
What's included:
- Technical specifications and build comparisons (index page)
- Tire clearance, BB standards, compatibility details (index page)
- Community feedback synthesis/trends, mainly from Chinertown threads (ratings page) with quality and reliability patterns I noticed
- EDIT: A two-tier ranking, between frames which seem to have overwhelmingly strong feedback and other frames, some with mixed feedback or too few entries
Limitations:
Some emerging frames like the Lightcarbon LCR018-D aren't included - first deliveries are just hitting the market now, so there isn't enough feedback yet. I've mentioned the new Tideace frame, not more. The selection reflects frames with more history or community discussion at the time I was researching. Some models like the LTK 268-D is “bundled” within the VB thread and has fewer entries but some of the mentions suggest it could be a notch better in terms of build. To call this dataset statistically significant would be wrong, but the level of knowledge and dedication on here and other forums makes the feedback highly credible in my opinion.
What's next:
I plan to add more frames in quarterly batch updates as feedback accumulates. If there's interest, I'd welcome community suggestions for which frames to research and add next. I’ve not considered Gravel frames or TT frames, just because I don’t look for any. I own my dream build gravel bike so road frames was top of mind, but if the site seems to deliver on its purpose- helping others in my shoes, I won’t exclude adding other categories over time.
No commercial interests:
No affiliate links or monetization - this is just my research notes made public. I've added placeholders in the code for Google ads in case I bother with that later, but no plans currently.
Check it out and let me know your thoughts. Constructive feedback welcome. This is an iterative process, meaning that I've published a MVP with the ambition to improve on it continuously.
bikeframeindex (dot) com
Edit/updates:
Actions taken regarding feedback below, which is very helpful in improving the site. Focus was on the "ratings" page where I've scrapped the points leading to the ranking and instead created two tiers which are neutral amongst the frames. Some minor tweaks and typos corrected.
-
I quite like it, but don't think you can blame the speed wobble someone experienced riding the Ican FL1 on the frame as it can be caused by a myriad of factors.
-
I quite like it, but don't think you can blame the speed wobble someone experienced riding the Ican FL1 on the frame as it can be caused by a myriad of factors.
Cheers. Thanks for the feedback, the speed wobble is perhaps a moot point. I recall someone else mentioning in the thread they had descended fast without similar issues. this is the kind of feedback that helps me refine the references.
-
Think I've hit 80+kph on mine a few times without issue ;)
-
How did the SPCycle R088 get a 2.8/5? Despite some build installation issues, it's been a solid frame overall. I'm not sure I can trust your rating index. You give frames with little data higher scores. Makes no sense.
-
Great idea, but I'd avoid assigning numbers because it only adds to confusion. Given QC can be all over the place, it's difficult to compare one to the next.
-
Are you using AI to read the posts and compile the summaries? Curious to know what model you are using because the summaries are awful.
Also there are a ton of mistakes here. For example, the LTK-266-D and the LTK266-D-SL are two different frames with different geometries, so the specs are wrong. Your summary also lists as a negative specific feedback I gave to someone regarding their potential fit with that frame. It is not a generalized negative at all. The newer ChatGPT models should be able to recognize context like this.
There are similar issues with the other frames as well. Serious issues like misaligned thru axles (on the top choice) and persistent headset issues are given the same space as..."carbon not being sanded down on the inside" when...well, not sure what sanding is ever done INSIDE a frame (the much lower rated Hygge Model S).
While the idea has potential, the current rating systems and site demonstrate a distinct lack of understanding of these frames and their price points, and are riddled with nonsense. Overall, it's a net negative to someone trying to understand the landscape of these frames.
-
How did the SPCycle R088 get a 2.8/5? Despite some build installation issues, it's been a solid frame overall. I'm not sure I can trust your rating index. You give frames with little data higher scores. Makes no sense.
Hi,
Thanks for the feedback. I know you're a big fan of that frame, and it's great it has worked out well for you. There have been some QC issues noted but perhaps i've been too harsh. I've decided to scrap the points all-together and just organize the frames into a top-tier and mixed-tier. Users can make up their own minds. Some things i've bookmarked are headset play (common amongst other frames too), this (which got replaced - but not great QC unless damaged in transit) https://chinertown.com/index.php/topic,5562.msg80652.html#msg80652 and then Tiny Dinosaurs headset issues.
Fair point regarding the frames with little data, hence my revision of tiers. I have assumed "OK until proven otherwise" and thence weighed some frames with more data points, but genuine issues, down more.
-
Great idea, but I'd avoid assigning numbers because it only adds to confusion. Given QC can be all over the place, it's difficult to compare one to the next.
Thank you for the feedback, and point taken. It helps me look at the site from different perspectives. I'll try to redo code for the site today and change it into a trusted category and mixed category which both are neutral as to the frames relative position. If you see other areas of improvement i'd love to know.
-
Are you using AI to read the posts and compile the summaries? Curious to know what model you are using because the summaries are awful.
Also there are a ton of mistakes here. For example, the LTK-266-D and the LTK266-D-SL are two different frames with different geometries, so the specs are wrong. Your summary also lists as a negative specific feedback I gave to someone regarding their potential fit with that frame. It is not a generalized negative at all. The newer ChatGPT models should be able to recognize context like this.
There are similar issues with the other frames as well. Serious issues like misaligned thru axles (on the top choice) and persistent headset issues are given the same space as..."carbon not being sanded down on the inside" when...well, not sure what sanding is ever done INSIDE a frame (the much lower rated Hygge Model S).
While the idea has potential, the current rating systems and site demonstrate a distinct lack of understanding of these frames and their price points, and are riddled with nonsense. Overall, it's a net negative to someone trying to understand the landscape of these frames.
Hi,
Thanks for the feedback. I am not using an AI model for this but reading the threads and taking notes/bookmarks myself. I have run a Python script (via beautiful soup) to scrape the threads into .txt files which makes reading a lot easier to do. Once I came across something noteworthy this approach allowed me to CTRL+F the entire thread given it's now in a combined .txt, in order to follow 1) the user/poster 2) topic by keyword(s) and look for a conclusion/data points.
The naming mistake will be corrected, I didn't add "-SL" with regard to that frame, but not sure what you mean by specs being wrong as it's BB86 and accepting 35c tires, as per what i've inputted on the website, or hwhat did I miss on this frame? I decided not to (as I also wrote on the site) categorize geometries or comment on them, since fit is highly individual; Regarding the drawback reference however, I got impressed by your level of knowledge & opinion on geometry (to even later on make references to 2010s Pinarello lines and some Liv bike) that I considered the fit question noteworthy. I've removed the comment, it doesn't belong there.
The 1056 for instance suggests consistently good user experiences to deserve a trusted rating in my book. The Hygge sanding reference was sloppy on my end, I (again, CTRL+F) looked up sanding and bookmarked the wrong comment & link for reference, when I had this https://chinertown.com/index.php/topic,4305.msg49861.html#msg49861 comment in mind, about having to sand the clamp due to slippage.
I've mainly worked on this website during short evening windows when the family go to sleep, and losing my train of thought begs for some mistakes along the way. Glad to get yours and others feedback to iron those out. My mindset in general is to publish work after a few iterations, and sort the issues along the way, rather than iterating forever. A few friendly people shared some feedback with me along the way which was very appreciated and helpful.
Whilst speculative buyers (my use case) have varying degrees of knowledge (I didn't even know B2C frame buying from China was a thing three months ago!) few people seem to be as invested and experienced as yourself, so you get the target audience. You probably don't need this kind of resource, whilst others might find some of it helpful on their journey. It takes a lot of time and effort to compare basics like BB standard, frame tolerance to reading through long threads to synthesize various data points into a summary. The initiative might never get your support (i came to think of this brilliant scene "tell me about your dream mosque?" https://www.instagram.com/reel/C11465SOSy3/?hl=en), but I 100% see the potential for a net positive, with continuous improvement.
Will try to make updates asap based on yours and others feedback.
-
I admire your ambition but I'm not confident about your scraping-based note taking system if you're missing the crucial facts (let alone the subjective pros and cons) of a frame like the LTK266-D-SL even now. @sazikashi has been regularly updating the original post in that thread, for instance, and it has all the information a website like yours should need. Even now you are mixing up the bb standards! BB86 is not T47. The biggest pro of the frame is the price-quality ratio and the WEIGHT which makes it the leader in its class. Cons might be aero-related for many buyers. Weight on hands? Irrelevant.
Im not sure this site is ready to be made public given the major errors for even just one recent open mould frame. If you're doing this for yourself that's great. But you did admit a lack of knowledge about bike buying and building that makes it hard for you to judge your own errors of commision and omission.
You're on to something with this database idea overall, and I like the design principles, but a quick glance based on my own knowledge of a particular frame suggests it's now more of a misinfo source than something 'opening up the black box' of china frames.
My suggestion is that for now the public facing site should be a list of popular frames, paired with industry-suppllied spec sheets, with links to their respective factory pages as well as their respective chinertown discussion threads. You should not editorialize from those threads: instead, prospective buyers should be encouraged to follow the link to read the thread themselves. It doesnt take so much time, and buying a frame is a major purchase decision anyways. Good luck with this project.
-
Thnx, love the work you (must have) punt in. Unfortunatly I'm not able to reach the site now. Getting a 'page not found'. Hope you're not discouraged by the slightly negative comments :)
-
Users can make up their own minds. Some things i've bookmarked are headset play (common amongst other frames too), this (which got replaced - but not great QC unless damaged in transit) https://chinertown.com/index.php/topic,5562.msg80652.html#msg80652 and then Tiny Dinosaurs headset issues.
At least SPCycle sent a replacement frame after verifying there was a structural issue with that particular setup. I remember a Winspace frameset had a giant void in the BB shell and ghosted the customer, while a TanTan X-68 frame had an exploding fork which could have resulted in actual death. If you want to get a better assessment of frames, just straight up read every thread. It's like relying on Cliff Notes as opposed to reading the actual book! ;)
-
At least SPCycle sent a replacement frame after verifying there was a structural issue with that particular setup. I remember a Winspace frameset had a giant void in the BB shell and ghosted the customer, while a TanTan X-68 frame had an exploding fork which could have resulted in actual death. If you want to get a better assessment of frames, just straight up read every thread. It's like relying on Cliff Notes as opposed to reading the actual book! ;)
Passed a lot of exams on cliff notes ;D
-
Hi,
Thanks for the feedback. I am not using an AI model for this but reading the threads and taking notes/bookmarks myself. I have run a Python script (via beautiful soup) to scrape the threads into .txt files which makes reading a lot easier to do. Once I came across something noteworthy this approach allowed me to CTRL+F the entire thread given it's now in a combined .txt, in order to follow 1) the user/poster 2) topic by keyword(s) and look for a conclusion/data points.
The naming mistake will be corrected, I didn't add "-SL" with regard to that frame, but not sure what you mean by specs being wrong as it's BB86 and accepting 35c tires, as per what i've inputted on the website, or hwhat did I miss on this frame?
The LTK266-D-SL frameset uses a T47i BB. The older frame uses a BB86. The 266-D-SL also has a different geometry than the 266-D, though confusingly, this is changing as they appear to be consolidating molds over there. Your scraping method has completely missed those details.
I don't mean to be a complete ass, but information like what you are putting together is only useful if it's accurate and includes relevant details. The challenge, though, is that it's tough to sort through the mess because little of it is straightforward.
For example, weights might be taken bare without hardware, bare with hardware, painted without hardware, or painted with hardware. Ultimately, what you care about is the painted, verified frame weight, but that's hard to find except from buyers here and often this is in photos of the bike on the scales.
Similarly, 35C is a statement of the max tire and wheel size the frame was designed for, but that does not indicate tire clearance. Rather, it's a naming convention that comes from 700c, 650b, etc, where [Nominal Wheel Diameter w/ tire]x[TireSize][Bead Seat Designation]. There are multiple tire clearance standards to be aware of. ISO4210 houses the major one of note. For road racing bikes, this is a minimum of 4mm clearance; for all other bikes, it's 6mm. When a manufacturer says its 35C tire, they mean the maximum tire size is 700x35C, but aren't stating which clearance standard they are using. 2mm on each side is a pretty big difference in room.
You should also consider whether the drawbacks of a frame can be easily resolved. A rubbing headset cap is annoying, but it's resolvable by a buyer; a badly misaligned thru axle is the kind of mistake that sends the bike to the trash. Additionally, some frames have standardized headtube dimensions and take regular round 56mm covers. Frames like these can be easily upgraded with higher-quality parts from major component suppliers. This is also an advantage for frames with UDH.
If you're determined to create a guide, also consider organizing the frames by price and use. Even within road bikes, endurance and performance geometries are distinct and meaningful functional differences. The easiest way to do this is to evaluate if a frame's geometry is closer to a Tarmac SL8/Giant TCR/Trek Emonda or a Roubaix/Giant Defy/Trek Domane.
-
I admire your ambition but I'm not confident about your scraping-based note taking system if you're missing the crucial facts (let alone the subjective pros and cons) of a frame like the LTK266-D-SL even now. @sazikashi has been regularly updating the original post in that thread, for instance, and it has all the information a website like yours should need. Even now you are mixing up the bb standards! BB86 is not T47. The biggest pro of the frame is the price-quality ratio and the WEIGHT which makes it the leader in its class. Cons might be aero-related for many buyers. Weight on hands? Irrelevant.
Im not sure this site is ready to be made public given the major errors for even just one recent open mould frame. If you're doing this for yourself that's great. But you did admit a lack of knowledge about bike buying and building that makes it hard for you to judge your own errors of commision and omission.
You're on to something with this database idea overall, and I like the design principles, but a quick glance based on my own knowledge of a particular frame suggests it's now more of a misinfo source than something 'opening up the black box' of china frames.
My suggestion is that for now the public facing site should be a list of popular frames, paired with industry-suppllied spec sheets, with links to their respective factory pages as well as their respective chinertown discussion threads. You should not editorialize from those threads: instead, prospective buyers should be encouraged to follow the link to read the thread themselves. It doesnt take so much time, and buying a frame is a major purchase decision anyways. Good luck with this project.
I appreciate the feedback, though you seem to be extrapolating from one frame you know well to dismiss the initiative. The site launched yesterday. Errors are being identified and fixed. If you find specific problems beyond the LTK266-D-SL, I'm grateful for feedback and I'll fix those too. But dismissing everything because of correctable mistakes or glitches this early is premature.
On the errors/typos: The BB standard typo is my fault - I can see how I originally logged it in my google spreadsheet as BB86, but it's been corrected today. Calling it wrong "even now" is misleading when the site has been public for less than 24 hours. The LTK266-D-SL entry was admittedly more rushed than others (I abandoned researching it early due to lack of large sizes for my own needs, though I did PM with Sakizashi to bounce some various impressions). These are fixable issues.
On your suggestion: Feel free to glance over the "index" page - it's pretty much what you're suggesting. Maybe you missed it? It lists frames I've followed with some high-level specs like BB type, build/mould method etc. I believe the table should be 95% correct (5% safety buffer inserted ;D). If someone wants to read hundreds of posts, they can (and probably should do when finalizing their shortlist!). The added value comes from consolidating patterns - my notes fast-track key observations. Adding too many floods the page, adding none...meh (I tried that...). Yes, I made mistakes on the 266-D-SL. That doesn't invalidate the approach.
On iterative development: Maybe you're unfamiliar with this concept, but it's a rather accepted way of working - continuous improvement based on feedback. Much like manufacturers featured here improve their frames over time. Here are two short summaries: https://entrepreneurship.hbs.edu/Documents/Session%20Summary/HBSRockMVPDevelopment.pdf and https://versions.com/topics/iterative-design/
"Rapid prototyping is a core part of iterative design. It emphasizes building just enough to test a hypothesis, not perfecting every detail from the start… Their value isn't in polish, but in speed to feedback. A rough idea, placed in front of users, can reveal more than a week of internal debate." (source: Versions.com link above)
-
Thnx, love the work you (must have) punt in. Unfortunatly I'm not able to reach the site now. Getting a 'page not found'. Hope you're not discouraged by the slightly negative comments :)
Thanks for checking it out - hopefully it'll load now in case you have a moment to try again. I did some updates and took it down for a while as I was swapping code. Would love to hear your feedback.
Re comments, this is not open heart surgery after all. It's an iterative process, and all feedback is helpful. Some input just motivates me to improve the initiative further. It's obviously not going to be perfect (ever), but definitely not 24h after launch. Hence the Who Is America reference I made before - some people will just never be convinced. If one person finds it helpful, I'm happy.
-
On your suggestion: Feel free to glance over the "index" page - it's pretty much what you're suggesting. Maybe you missed it? It lists frames I've followed with some high-level specs like BB type, build/mould method etc. I believe the table should be 95% correct (5% safety buffer inserted ;D). If someone wants to read hundreds of posts, they can (and probably should do when finalizing their shortlist!). The added value comes from consolidating patterns - my notes fast-track key observations. Adding too many floods the page, adding none...meh (I tried that...). Yes, I made mistakes on the 266-D-SL. That doesn't invalidate the approach.
ack. A rough idea, placed in front of users, can reveal more than a week of internal debate."[/i] (source: Versions.com link above)
There is a need for what you are trying to make, so the goal is good, but the current product you've released does nothing to validate your approach. Rather, it suggests that you are consolidating the wrong patterns, prioritizing the wrong information to summarize, and using a method that leads to the omission of important information.
-
I wasn't going to comment, but now I am.
Your work on this list will end up being short lived. The information you generate will be quickly sucked into the realm of the internet and IA processing such as ChatGPT. I used to make Excel spreadsheets / lists of frames and components trying to determine what frame would best fit my needs. More and more, I just have ChatGTP do the general research and then fine tune it. I input the frame brand, name, and ask about its geometry and available specifications. Then I ask it to go to "bike Insights" to find all of the other frames with almost the same geometry. One can quickly refine questions and get great results. (For Chinese frames, many of the details are taken from Chinertown and thread postings concerning specific frames. )
There is another recent thread about making a frame list where the OP uses ChatGTP to help him make his list.
ChatGTP can and does search through all of the public sites gleaning information about what you ask. It is a whole new world...
I am torn about letting AI tools do the work, versus the pleasure/satisfaction I used to like doing research on "bicycles" OTOH, I am getting into the last 1/3 of my life, and I'de rather be riding than researching bike frames/parts, therefore ChatGTP has been very enabling to me.
-
I appreciate the feedback, though you seem to be extrapolating from one frame you know well to dismiss the initiative. The site launched yesterday. Errors are being identified and fixed. If you find specific problems beyond the LTK266-D-SL, I'm grateful for feedback and I'll fix those too. But dismissing everything because of correctable mistakes or glitches this early is premature.
On the errors/typos: The BB standard typo is my fault - I can see how I originally logged it in my google spreadsheet as BB86, but it's been corrected today. Calling it wrong "even now" is misleading when the site has been public for less than 24 hours. The LTK266-D-SL entry was admittedly more rushed than others (I abandoned researching it early due to lack of large sizes for my own needs, though I did PM with Sakizashi to bounce some various impressions). These are fixable issues.
On your suggestion: Feel free to glance over the "index" page - it's pretty much what you're suggesting. Maybe you missed it? It lists frames I've followed with some high-level specs like BB type, build/mould method etc. I believe the table should be 95% correct (5% safety buffer inserted ;D). If someone wants to read hundreds of posts, they can (and probably should do when finalizing their shortlist!). The added value comes from consolidating patterns - my notes fast-track key observations. Adding too many floods the page, adding none...meh (I tried that...). Yes, I made mistakes on the 266-D-SL. That doesn't invalidate the approach.
On iterative development: Maybe you're unfamiliar with this concept, but it's a rather accepted way of working - continuous improvement based on feedback. Much like manufacturers featured here improve their frames over time. Here are two short summaries: https://entrepreneurship.hbs.edu/Documents/Session%20Summary/HBSRockMVPDevelopment.pdf and https://versions.com/topics/iterative-design/
"Rapid prototyping is a core part of iterative design. It emphasizes building just enough to test a hypothesis, not perfecting every detail from the start… Their value isn't in polish, but in speed to feedback. A rough idea, placed in front of users, can reveal more than a week of internal debate." (source: Versions.com link above)
1) I made pretty clear that I was using the one frame thread I knew by heart as an example--so yeah, definitely and purposefully extrapolating. But what I found was terribly wanting. Incorrect or beside the point. It didn't bode well for the other information aggregation. I wouldn't say I'm dismissing everything, by any means. I like the concept.
2) No need for the pedantry on iterative development. The problem is not with any of the mechanics of the site itself--the engineering part--but rather with your human-all-too-human curation process and the faults therein. YOU are the one doing the identification of pros/cons, what is important/unimportant to buyers, and so on. It's not the coding/development of the website that is lacking. It's in another kind of expertise, of which a more patient approach to compiling documentation, personal anecdotes, and experiential knowledge from the threads would be the guarantor of a quality, public-facing compendium of our hard-won collective knowledge and expertise.
Not dismissing this at all -- it's really cool and the site is slick. But making it trustworthy is going to require a more careful approach, in my opinion. Keep at it.
-
I wasn't going to comment, but now I am.
Your work on this list will end up being short lived. The information you generate will be quickly sucked into the realm of the internet and IA processing such as ChatGPT. I used to make Excel spreadsheets / lists of frames and components trying to determine what frame would best fit my needs. More and more, I just have ChatGTP do the general research and then fine tune it. I input the frame brand, name, and ask about its geometry and available specifications. Then I ask it to go to "bike Insights" to find all of the other frames with almost the same geometry. One can quickly refine questions and get great results. (For Chinese frames, many of the details are taken from Chinertown and thread postings concerning specific frames. )
There is another recent thread about making a frame list where the OP uses ChatGTP to help him make his list.
ChatGTP can and does search through all of the public sites gleaning information about what you ask. It is a whole new world...
I am torn about letting AI tools do the work, versus the pleasure/satisfaction I used to like doing research on "bicycles" OTOH, I am getting into the last 1/3 of my life, and I'de rather be riding than researching bike frames/parts, therefore ChatGTP has been very enabling to me.
Very interesting reflections, thanks for sharing. There's always a personal trade-off to be made and I think you seem to be prioritizing your time well by focusing on riding your bike!
In terms of AI, some of the models might also end up being short lived. Here's a fitting article, if you see beyond the SEO layer, it becomes evident that these limitations will restrict various AI models significantly - they just won't reach far enough and Google is the internet. Enter Gemini.
https://www.webpronews.com/googles-10-result-limit-disrupts-seo-tracking-tools-and-inflates-costs/
-
I like this concept. I just have two requests:
1) Would you be able to create a category for disc and rim brakes? Yes, I know pretty much every frame today are running discs, but I just enjoy rim brakes for no particular reason lol.
2) Would you be able to make it community run? Like Wikipedia on how users can edit pages, maybe you could make it community-ran like here. That way, the information presented is more accurate and ideally, contributed by first hand accounts. To prevent trolls, maybe you could give editing permissions to specific users that request for it?
-
I like this concept. I just have two requests:
1) Would you be able to create a category for disc and rim brakes? Yes, I know pretty much every frame today are running discs, but I just enjoy rim brakes for no particular reason lol.
2) Would you be able to make it community run? Like Wikipedia on how users can edit pages, maybe you could make it community-ran like here. That way, the information presented is more accurate and ideally, contributed by first hand accounts. To prevent trolls, maybe you could give editing permissions to specific users that request for it?
Cool feedback, thanks for sharing.
1) Of course it would be possible, and frankly quite simple for me to do under the "index" table. Very few rim brake builds featured on here (more on Weight Weenies), but there's lots of them amongst various suppliers (example: https://www.ltbikes.com/index/products/index/cid/18.html)
2) Yes, that is an idea. Could be done similarly to how the contests here are run (like the recent Tideace frame), meaning a contributor would need for example a minimum number of posts, time on the forum or whatever is a reasonable denominator. Votes could be done by means of "liking" their contribution. I am completely open minded :)
If you, or anyone would like to share a particular feature or even make a blog post i'd be happy to publish it.
-
1) I made pretty clear that I was using the one frame thread I knew by heart as an example--so yeah, definitely and purposefully extrapolating. But what I found was terribly wanting. Incorrect or beside the point. It didn't bode well for the other information aggregation. I wouldn't say I'm dismissing everything, by any means. I like the concept.
2) No need for the pedantry on iterative development. The problem is not with any of the mechanics of the site itself--the engineering part--but rather with your human-all-too-human curation process and the faults therein. YOU are the one doing the identification of pros/cons, what is important/unimportant to buyers, and so on. It's not the coding/development of the website that is lacking. It's in another kind of expertise, of which a more patient approach to compiling documentation, personal anecdotes, and experiential knowledge from the threads would be the guarantor of a quality, public-facing compendium of our hard-won collective knowledge and expertise.
Not dismissing this at all -- it's really cool and the site is slick. But making it trustworthy is going to require a more careful approach, in my opinion. Keep at it.
I read your points with great interest and am sure some improvements can be made, but this will also never become a work on par with Robert Caro and the site won’t satisfy your needs, that’s OK.
Re iterative development, the concept extends beyond code, by which I mean that content can be amended (documentation, personal anecdotes and the experiential knowledge you have in mind). Some of the comments are illustrative but will never be exhaustive or the best representation of absolute pros/cons to everyone. I’ve aimed for some (sometimes trending ones) to trigger thoughts like “ah, I’ll need to consider possible headset issues with this frame, let’s dig into that before I make my order”.
-
This makes me begin to wonder if one of the "greater minds" on this forum should start a thread on how to research Chinese bicycle frame/ parts using AI tools. If we all band together, we could create a standard SOP that we agree on to compile a frame and/or component database.
One example I learned 5-6 months ago was this generic Google search method/syntax;
site:chinertown.com "your search terms here"
(Since Chinertown's search feature is limited, I have used this many times to go back and research postings on Chinertown.)
I have not found Google's top ten limitation as being a hindrance in my recent searches.
Here is an example of what I will search out: I will open several new windows to search on 2 different frames, say Carbonda or Speeder
In each window I will type a specific website search
site:chinertown.com geometry compare Carbonda 505 vs 707
site:forums.mtbr.com geometry compare Carbonda 505 vs 707
Site: (weightweenies)
Site: (Reddit)
Site: (Pinkbike)
(you get the gist)
I look at the results on each page to see what forum specific results get displayed.
Then I will do similar searches for price, quality, and other things on my mind about the frame/s.
When I get a general feel for what I am researching and a general perspective, I will write up a more specific query to ChatGTP asking it to provide me with specific information, or details where the response ends up being generally very informational.
I am aware that garbage-in often means garbage-out, so I end up on a "learning curve" iterating with additional questions in order to get back information I am willing to believe, or can confirm as good data.
-I have found that when I do initial research using the Google search engine, ChatGTP will pull from all the sites I visited at what appears to be a higher priority versus, it doing its own "from scratch" search.
-
One other piece of a "bike ranking/rating" list is that many of the complaints or compliments regarding a certain frame manufacturer/model is based on an individual's forum posting, and most often, there is not data provided to substantiate the claim, so to me it becomes an "opinion" and not fact based.
Perhaps, if we actually go down a frame/component rabbit hole, we create a thread that lists specific problems in a format, that our forum members detail the specific problem, and its solution/resolution. This list will grow over time, and an AI tool could read and evaluate this list, and populate the frame/component database.