I get extremely frustrated when I have to manually crawl through a shop system just to understand their naming scheme and learn what the technical specs and differences between each product are.
This is hard and annoying enough with a single manufacturer like Convoy, but it gets really frustrating very quickly if you want to compare between brands.
Especially annoying to me is the Firefly Lights site.
Just give me a list with all your products and their specifications!
For this, I have always loved parametric searches like the ones from Mouser, DigiKey, or LCSC.
M Maybe I’m the only one who craves this sort of list… but if I’m not, this would be a big team effort: compiling some sort of JSON containing all the relevant information you want to sort by, and even figuring out what those parameters should be in the first place.
So what would you want to have as parameters in a parametric search?
Maybe this could also be cross-referenced and linked to reviews, lists, and other sites, like for example HCRI:
Not functional yet — consider this more of an early draft/proof of concept:
So now I’m interested in what you think about this idea.
The Parametrek website is essentially what you’re after, but it’s not up-to-date because it’s an enormous task.
I don’t disagree that this would be useful, I just can’t see how anyone can maintain the database as a hobby. I completely share your frustration with naming conventions, I feel like at least with logical naming convention, one could at least have a good guess at what each model is!
On the convoy site alone, the number of variations is enormous just S2+ lights:
25 host colours/materials
~70 LEDs accounting for CCT options.
5+ drivers, obviously not all work with LED options.
Let’s assume average of 2 driver options per host/led combo… that’s 3500 different SKUs
This is why it must be a some what distributed task.
Tools can help a lot to scrap a site. and scraper tools can be “Ai” coded. the resulting data can also parsed by Ai…
seams to work some what but need some amount of human supervision.
Also to add, yes I suspect at least an attempt could be made to scrape review websites for details for LLM processing, however, how do you compensate the reviewers for their data?
Yup, I built a super review site that pulled in information from product reviewers, manufacturer web sites as well as some synthetic details that would be consistent from one product to the next (beam shape, etc.).
Where I got hung up is that although I believe it would have been fair use (I was linking to the review web sites and pulling out attributed quotes), I didn’t want to offend anyone. So for now it’s just private for me lol.
At least one reviewer on here has complained of manufacturers “profiteering” from their test results, I’m not sure I see the difference if you were to make this public, I’m afraid.
What worries me is that at some point, the quality reviewer will give up their goodwill because it’ll just be scraped and repackaged and not attributed.
At that point we will only be left with cr@p “reviewers” using AI to generate a bunch of fluff to post, then the bots will scrape the AI slop and we will be left with endless drivel to wade through with no idea what to trust.
I think the natural human dopamine reward “ah I know this answer!” Is what currently drives most posts/suggestions when users post a reccomendation thread- there’s something very human about being able to say “This light matches most of your requirements but in my experience, the button isn’t very satisfying” etc. I fear in the coming years, people will be recommended the perfect light as per their requirements, but the human forum interaction will be dead.
Look up Foy’s reviews threads for the “vibe” of what I consider is going to be lost… Maybe this is just me getting old though.
parametric search for flashlights is one of those things that sounds obvious but nobody has done it well. paramatrek was the closest attempt but the database fell behind. the hard part is not the search UI — it is keeping the data accurate when new models drop every week and specs change between batches. if you are building something, i would suggest focusing on the few params people actually filter by: cell type, switch location, emitter options (at least CCT and CRI), and beam profile. everything else is noise for 90% of buyers. happy to help with the LED/emitter data side if you need it.
I agree completely. I think there are roughly 4 categories of flashlight reviews. First - people who take some pictures and largely regurgitate manufacturers specs. I find these pretty much useless and they’re far too common. They might throw a bit of their own judgment, but it’s often not very useful. Second, people who do genuine testing - spectrometers, lumen tubes, runtimes, etc. Those can be really helpful. Third, people who have a unique perspective or focus, or who give useful and interesting feedback about how a light operates or functions. Depending on how interesting their perspective is, they can be great. I think this is probably the kind you were referring to. Fourth category would be some combo of the other 3.
I certainly don’t want to discourage people from engaging in genuine reviews. Even the useless reviews don’t bother me - I just don’t get anything out of them.
