Show HN: StumblingOn – Randomly find indie websites – recreating StumbleUpon | Hacker News Reader

Show HN: StumblingOn – Randomly find indie websites – recreating StumbleUpon

stumblingon.com

138 pointsALittleLight61 comments

Needs a way to flag abuse, like https://get portal.co/earlyaccess which is just some company's product with a slick sales site. First thing I got when I clicked :(

Edit: another abusive link https://www.the followershop.com/ (i have broken the links to avoid any sort of SEO for them)

Good point. I have a "like" button in the history section. I will add a "report"/"bad"/"shouldn't be here" button to that as well.

Oh hey, you're the creator! This is a really cool site, thank you.

Thank you!

Oh also for the sites marked down as bad, create a public /shame list that’s available as plain text/json and we can use it to add to block lists on our own projects.

Not a terrible idea but that could be abused just as badly. I sure hope there's a way to get yourself off this list if you are falsely accused.

I had thought about adding a "top ten likes" list, but decided against it because people might try to like their own website to get it highly ranked, and it would be very easy to programmatically like a website an arbitrary number of times. I think showing a list of bad websites has a similar problem, e.g. there is a communist website in there and some kind of polyamorous Christian-swingers forum, if someone had ideological problems with a website like this they could try to flag it.

Rather than sharing feedback, which might create bad incentives, I'll just add it to my review page, where I review and approve submissions. I'll look and see if anything is getting reported and then choose to remove or leave it.

Hey, just wanted to say the first site I got was https://www.the-odin.com, and I thought it was particularly random and cool. "Genetically Engineer Any Brewing or Baking Yeast to Fluoresce" was not at the top of my list for cool presents, but I thought it was super cool!

There is a decent amount of content on Netflix covering the man behind the website. He injected himself with some protein that was supposed to make him more muscular in public as a publicity stunt. Pretty controversial but also pretty interesting.

I made something similar but a bit more random.

https://randomdomain.benbristow.co.uk

Keeps taking a random word from an English dictionary and a random popular domain extension (com/net/org etc.), pings it and returns the first one that responds.

Quite fun once you sift through the parking pages. I need to work on something to detect those.

Ah, now that really is random! One idea for sifting parking pages that comes to my mind is using image recognition. I can easily tell when something is a parking page by looking at it, I bet you could get an ML model to do so with pretty good accuracy too.

There is a class called fast.ai [1] which I personally found to be an amazing introduction to machine learning / deep learning. The first hour and a half course gets you using their library to make an image classifier that can differentiate dog and cat breeds. I bet by simply gathering training data (sites that are and are not parking pages) and reusing the same architecture used in that first class you could train a model that does a pretty good job of it.

Once you could classify pages you could just pre-compute a large list, classify, throwaway all the parking pages, and make the remainder available to your user.

1 - https://www.fast.ai/

Haha, cheers.

That sounds pretty fun.

Knowing myself though I'd just make a manual checker as most of the parking sites are ran by a few companies (e.g. GoDaddy etc.) so they'd be pretty easy to pick-up.

That plus a keyword search for words like parked, parking, buy now, etc. Might have similar effectiveness!

Cool idea but the first 3 websites I get are all parking sites.

I'm pretty sure with a simple keyword check, you can eliminate most of them. A small amount of fasle positives are ok in this case I guess.

This is great! I was just thinking about StumbleUpon the other day and wishing it were still a thing. I have a whole folder full of awesome bookmarks (half of which are dead now) I found using StumbleUpon.

Thanks. A while back I was trying to find the original StumbleUpon and discovered it had become some kind of social media site, so I recreated what I remembered was the core functionality (click button, get random site).

> A while back I was trying to find the original StumbleUpon and discovered it had become some kind of social media site

I don't know what StumbleUpon is like now, but it was always a social media site; a major feature was the stumpleupon comments page associated with any actual page.

One of the things you could do with the original StumbleUpon is stumble by category. I had categories for my hobbies and it was really fun to stumble on different pages relating to one of my hobbies. As it is, this is fun and takes you to sites you'd never have gone to on your own!

Another idea - people have already mentioned the ability to report pages but I think it'd be cool to be able to upvote a page. Then for each category you could have a rolling leaderboard of the most upvoted pages in the past 7 days or whatever. Some people might just want to look at that, or look at that in addition to getting rando pages.

Good luck!

Thank you. Categories are a good idea. One category I've thought about is videos as a few people submit those and I've been rejecting them because they don't seem like webpages. Putting them in their own category might solve that though.

As far as upvotes, there is currently a heart button you can click next to the website on the history section. Not very discoverable (looking at the metrics about 1 stumble in a thousand is getting a heart).

Right now the hearts don't do anything but serve as a reminder for a user that they liked a page. I've thought about possibilities to incorporate that into the algorithm (e.g. more likely to get websites that have higher like rates) and having a leaderboard is a good idea too, though I worry that it would give an incentive to manipulate the like buttons. Creating a new user right now is as easy as just generating a random string, so a bad actor could easily cause infinite likes to a page.

If you could implement categories, I think this could be almost perfect. I've found that's what I really miss about StumbleUpon. Being able to randomly stumble upon things from a variety of categories that interested me was really great. Found plenty of awesome things through there my freshmen year of college.

I also made something similar called Stumblefun[0], trying to recreate that StumbleUpon experience but without the need to install a toolbar.

Right now it's only focused on SaaS products you may find useful but I'd like to add other categories of websites you can discover in the future.

[0]: https://stumblefun.com/

Woah this is cocoocoocoocoool

Thanks! I know there are some old urls in there that no longer load successfully that I need to clean up. It's been a few months since I've done any work on this but maybe this is some good motivation to get back to it :)

I really like it too - but one that stumped me when I landed in the homepage was "What do I click to see something?" Maybe add a big green "Ready, let's go!" button underneath the instructions/controls.

I like the concept, and I like having that the webpage is loaded in a frame so you can keep your controls on top. One thing that might make it easier/faster is if you could use the keyboard controls to go right, up, etc.

The in-frame load broke it for me. None of them loaded in on Firefox 84, and some explicitly showed an error saying they weren't allowed to load as an embedded frame. I had to open each site in a new tab to be able to see what it was. Loved the service despite that, tbh.

I am only getting various product websites. Is that intentional?

Thank you very much. Like a lot of people here, SU was a big part of my internet exploration as a child. This has already given me lots of cool pages to explore!

Just like the original StumbleUpon, within a few clicks, I discovered several fascinating sites. Thanks for posting this, it is a real treasure.

Thank you. I think it is really hard nowadays to find small independent websites. Just a couple days ago I remember reading a hackernews comment from a person saying that most of their web traffic started from just a few places (hackernews, reddit, twitter, etc) - and that really resonated with me. I hope that something like stumblingon (or a better solution) can help people find unique and interesting content.

Very nice! Almost all of the traffic to https://datacrayon.com comes from HN/Reddit... I remember things being different in the GeoCities/Angelfire days. I wonder if sites like this would change that

For sure - I am really interested in finding ways that would help new or smaller websites get traffic and in web discovery.

I also wonder if Google is somehow getting worse about this. For example, I happen to know now (though I didn't know when I started the stumblingon project) that there are a couple other similar sites that exist. However, when I originally googled "Stumble Upon" and phrases like "Stumble Upon Alternative" or "Stumble Upon Clone" I never saw any of the others that do exist. Instead, it is/was mostly "Top N Stumble Upon Alternatives"-type listicles that weren't very good and would suggest things like "Facebook". Oddly, the top result, for me, for "Stumble Upon Clone" is this weird website [1] which I think is selling Stumble Upon clones, like - they will make a Stumble Upon clone for you. (Possible I'm misreading that).

1- https://www.ncrypted.net/stumbleupon-clone

Thank you! I love exploring the little corners of the internet, but haven't been exploring around as much in the past 10 years.

I really appreciate the History portion. If I recall right, StumbleUpon with the toolbar didn't save to history quite right, so I had a tough time getting back to a page after getting click happy. Down the line, liking/categorizing/personalized suggestions would be really nice, but I might be too reminiscing about StumbleUpon too much. :)

I remember setting the Stumble button to Alt+F11. Those two keys were smooth without any markings remaining after a couple years.

The Stumble button was hidden at my screensize which was a little confusing :-). Looks like removing `.wholePageContainer{height: 100vh;}` fixes the issue and keeps the buttons from stacking on top of each other.

Hmm, that's troubling. What browser and OS are you using? Having read your comments I realized I haven't tested it on high zoom in safari (or at all on Internet Explorer or Edge).

Safari on Mac, for example at 950x950, shows them stacking. Looks like Firefox on Mac is ok.

Interesting. Thanks for letting me know about this. I'll get it fixed this evening.

Sure thing and thanks for bringing back good StumbleUpon memories!

Gosh I shudder thinking about how many countless malicious websites I visited in high school using StumbleUpon...

Hopefully there are zero malicious websites here, but the metrics page shows a little over 10k stumbles since submission - https://service.stumblingon.com/metrics

The first site I stumbled on from this project offered to sell me instagram likes for $$.

Thank you! This is BIG!

And this API endpoint is very interesting:

  curl -X POST "https://service.stumblingon.com/getSubmissions" -H  "accept: application/json" -H  "Content-Type: application/json" -d "{\"status\":0}"