Hacker News Reader

Comment by semolino | Hacker News Reader

The Internet's been weird again. Cool, fresh stuff has always been available on the web. The author fails to address the inherent problem with obscure personal/artist sites (discoverability), and neglects to mention the best way to keep up with niche projects (RSS). Until use of personal aggregators becomes widespread again, users will opt for the default, algorithmically-driven drip that prioritizes popular content masking itself as "relevant to you". You can lead a horse to water, but you can't compel it to mindfully curate its media intake.

View on HN

I think considering RSS as the solution ignores the upsides that algorithms do have, that of analyzing behavior, it's almost interactive.

The problem of course is what the algorithms are generally tuned towards. Imagine instead a sort of open source algorithm where you're in control of how you want to tune your goals.

I often dream of this, the highest technology with in depth behavior analysis, but in service of each individual user instead of the tech overlords and its army of advertisers.

That's basically what StumbleUpon was, and it was great while it lasted. You clicked like on stuff you liked, and it would show you more stuff that other people who liked similar things liked. If someone could make something like that again for 2023, I would use it.

There's also the aspect of timing, an algorithm could pace different styles of content and analyze how the themes and purpose keep you engaged, in which order to show them, as well as what's the best pacing of longer and shorter content in order to have the best experience.

StumbleUpon failed to monetize in a way that let it improve its algorithms and resources, enough to stay relevant to (most) people.

The most obvious thing would have been advertising -> enshittification. Otherwise, they needed some other form of profit -> improvement -> more profit solution.

The economics of these systems is a first level problem, inextricable from the good they do.

For me that is the answer as to why everything doesn't run through Tor. You can't scale for demand if demand doesn't reliably contribute at least as much value to its own supply quantity and quality.

I think what you're saying is definitely a factor, but I like to imagine it's a factor for innovation, but once the capability has been breached, then the possibilities for implementing something similar on a smaller scale, and thus possibly divorced from ad revenue, then we could see algorithms implemented with goals set by individuals for their own good.

Kind of like how now there's not just reddit clones all over, but you can basically spin your own.

I think there is a convergence in our views.

To work for our personal interests, and only us (i.e. no third party), we need to pay for the resources directly, whether its hardware or a cloud service. Whether we contribute to an open algorithm or pay others to work on the algorithm.

We can certainly provide our own hardware and innovation.

But most people don't want to manage these things themselves, they want a service.

Another reason for the cloud/provider solution, is that a lot of these services have functions that are amortizable across users: i.e. crawling, indexing, correlation analysis, etc., that precede making individual recommendations.

Either way, if people don't contribute in proportion to their demand for service quality, scale and innovation, the system is going to either die, or get subsidized by third parties who find the resources to be highly competitive by using us.

I don't follow why it can't be software in the same way that you can download blender and make your own animated movie if you have a beefy enough computer.

That is what I meant with:

> We can certainly provide our own hardware and innovation.

I.e. our own hardware for running software.

Most people in the western world have way more computing power (and memory, and network bandwidth) than they would ever need. We as an industry have chosen to "borrow" that extra capacity through advertising and bitcoin mining and insanely wasteful webapps, but there's no reason we can't develop shared software, where the crawling, indexing etc all happens distributed - if StumbleUpon told me it will use 10% of my CPU to parse and index zero-knowledge encrypted data from other users, I would consider that a very good deal

Agreed; hopefully it will soon be easy to deploy intelligent, locally-run/configurable user agents to crawl the web on our behalf and find our personal gold.

I still think RSS is a great way of encapsulating data in a format that's almost equally human- and machine-readable, regardless of how "smart" you want your aggregation to be.

I wish I had a use for RSS, but the few websites I actively read, I enjoy their design, not just their content, and the rest is on YT.

Maybe I lack imagination in how RSS could help me beyond being a collection of articles unified under a common GUI.

If you want to provide details on how you use it I'd be glad to read it.

You can still use it to get updates from websites you like reading, and then click through to the full page on the website without reading the content inside your feed reader.

Who would pay for it? That’s always the problem with nice things in software and online. Advertisers pay for the lobotomy engines that drive social media.

Nowadays one can spin their own email service or media server. Why can't we see something similar happening in the near future with AI assistants? I don't mean universally accessible, it might only be for tech savvy people.

I'm going to take a moment to pitch an experimental service / pet project on my back burner. I call it "the hourly drip". Its a content feed, but you have to pay to post content. I know that sounds like a feed of pure advertisement but here's the catch: the price of posting is algorithmically adjusted to keep the rate of new posts fixed at an average of 1 per hour. In other words, the pricing mechanism isn't about profit (in fact, its bad at maximizing profits). Rather, its a form of quality filter that reduces the feed down to a digestible amount of content per day.

what one should be using nowadays to discover cool fresh stuff?

Reply on news.ycombinator.com