7 ms·

I like the idea of this. There's so much information on the web, but we still need a way to bring that information to other applications, without being tied to

I like the idea of this. There's so much information on the web, but we still need a way to bring that information to other applications, without being tied to a particular source. That was really the dream of the semantic web, after all. This kind of idea would be really nicely paired with good Microformats[1] support, which continues to be a very good idea. That way we can find, say, a recipe or an address on a web page in a reusable way and without needing magical heuristics. (Of course, "reusable" in theory, with the caveat that everybody forgot about microformats around when Google decided they could machine learn their way out of everything). [1] http://microformats.org http://microformats.org

MaxBarraclough 5y ago
> I like the idea of this. There's so much information on the web, but we still need a way to bring that information to other applications, without being tied to a particular source. I'm not sure I'm interpreting you correctly here, but I think I'm on the other side of this. The problem is that many modern websites are godawful. I think the story pretty much ends there. If websites were not awful, we wouldn't find ourselves appalled by the idea of just embedding a browser. Modern web browsers feature a 'reader mode' as a countermeasure to that much modern web design is significantly worse than having no web design at all. If you're serious about a 'lightweight' alternative to the lumbering horror-show of the modern web, the way forward is either Gemini [0], or a formalised simple subset of HTML. [1] > That way we can find, say, a recipe or an address on a web page in a reusable way and without needing magical heuristics. I think the find and reusable aspects here are really two very different problems. The reusable part is easy. HTML is already reusable. A standardised simple subset would be even more so. [1] The find part is trickier. Discovering decent content is harder, as there's an arms race of ad-funded spammers trying to out-compete legitimate recipe sites in search-engine rankings. (There's also the possibility of search engines not being motivated to work on delivering good search results. [2]) The idea of having a choice between native GUI applications and web apps, has been with us for some time. Email is probably the best example, we've long had the choice between webmail and native email clients. Beyond webmail, these days even Microsoft Word has a web-based version. There are of course both advantages and disadvantages to web-based applications. [0] https://news.ycombinator.com/item?id=23730408 https://news.ycombinator.com/item?id=23730408 [1] https://news.ycombinator.com/item?id=29291392 https://news.ycombinator.com/item?id=29291392 [2] https://news.ycombinator.com/item?id=29772136 https://news.ycombinator.com/item?id=29772136
- Karrot_Kream 5y ago
  How do you know someone uses Gemini? They'll tell you the moment they can! Like the vegans of the web...
  - antifa 5y ago
    I didn't want to register for a gemini group just to ask this, but I looked at it from a webdev point of view just long enough to notice there was no evidence of any way to implement a "save comment" or "type in search" feature, basically as if gemini was entirely reduced to mostly static websites with no forms of any kind possible. Is that an accurate assessment?
    - Karrot_Kream 5y ago
      > I didn't want to register for a gemini group just to ask this You actually don't need to. Gemini is a client-server architecture just like the web. You can grab a Gemini client, or use some of the web portals out there that make a server-side Gemini request and render the result to the browser. > or "type in search" feature There is a limited form of input allowed, but this input shows up as a query parameter essentially. This is how you can do search or offer some form of interactivity and how the current Gemini search engines/crawlers work. > basically as if gemini was entirely reduced to mostly static websites with no forms of any kind possible Again Gemini does allow limited forms, but it only accepts a single query parameter as input. You can, of course, parse the parameter however you choose, but the general culture is that there's very little interactivity and most everything is a static page.
- jka 5y ago
  Also entirely possible that I'm misunderstanding both of you, but I think what the parent comment was imagining was something like common schemas (like schema.org[1]?) for content that is currently around the web and encased in the challenging web design you mention. With common (and evolving) formats -- and incentives for publishers to provide their information within those formats -- we could then have much simpler, more streamlined tools to use and remix that data in application-specific ways. [1] - https://schema.org/docs/full.html https://schema.org/docs/full.html
hombre_fatal 5y ago
One problem is that it takes a lot of work and effort to build any of the valuable hubs where people post information. Ever try to start a forum? It's a monumental task with no guarantee of success. You may even need to employ people to grow and maintain one. And once you've finally grown one of these hubs that accumulates recipes, lyrics, real estate listings, classifieds, etc. (whatever you had in mind) there's no incentive to make it as easy as possible to share it with the world. Once you get over "ugh, everyone just wants to make a buck", there's the fact that it wasn't free to build and maintain the platform to begin with. And perhaps the only incentive to build the platform was the idea that people would pay for the value. Or, who is supposed to do the work of curating and organizing all of this information and then producing an API so that others can build on it, and why haven't they started? There are probably some inconvenient truths in the answer beyond cynicism.
- anderspitman 5y ago
  Ha, I'm trying to start a data ownership forum now. My approach has been to have it be a central place for support for all my open source projects. We'll see how that works out.
  - antifa 5y ago
    > data ownership forum Care to share or explain more? Sounds awesome.
    - anderspitman 5y ago
      That's pretty much the whole tweet! Feel free to join us: https://forum.indiebits.io/ https://forum.indiebits.io/
dillondoyle 5y ago
Funnily, Google has pushed websites to add more structured data into their html for crawling.. Seems it's used for SEO hackinge. For instance on recipes why wouldn't a site give their recipe a super high rating. Those sites are awful SEO spam adservers basically. But business info that can be used to map seems pretty valuable to google. https://developers.google.com/search/docs/advanced/structured-data/search-gallery https://developers.google.com/search/docs/advanced/structure...
- tweetle_beetle 5y ago
  It's not as easy as faking metadata of 5* user ratings of your own content. If it ever was, that technique is no longer useful and likely to lead to losing rank. Google uses a subset of schema.org, validates it according to its own stricter specification and extends it in other ways. It doesn't naively consume everything you provide. For example [1] > If the Recipe structured data contains a single review, the reviewer's name must be a valid person or organization. For example, "50% off ingredients" is not a valid name for a reviewer. > Warning: If your site violates one or more of these guidelines, then Google may take manual action against it. Once you have remedied the problem, you can submit your site for reconsideration. [1] https://developers.google.com/search/docs/advanced/structured-data/review-snippet#guidelines https://developers.google.com/search/docs/advanced/structure...
  - dillondoyle 5y ago
    Makes sense. And I'm sure they don't want to give too much away just helps SEO spammers even more
Mandatum 5y ago
I love RSS and MQTT for this as a "good enough" method. There's a few QL choices for CRUD operations too. We're getting there, albeit slowly..

MaxBarraclough 5y ago

> I like the idea of this. There's so much information on the web, but we still need a way to bring that information to other applications, without being tied to a particular source. I'm not sure I'm interpreting you correctly here, but I think I'm on the other side of this. The problem is that many modern websites are godawful. I think the story pretty much ends there. If websites were not awful, we wouldn't find ourselves appalled by the idea of just embedding a browser. Modern web browsers feature a 'reader mode' as a countermeasure to that much modern web design is significantly worse than having no web design at all. If you're serious about a 'lightweight' alternative to the lumbering horror-show of the modern web, the way forward is either Gemini [0], or a formalised simple subset of HTML. [1] > That way we can find, say, a recipe or an address on a web page in a reusable way and without needing magical heuristics. I think the find and reusable aspects here are really two very different problems. The reusable part is easy. HTML is already reusable. A standardised simple subset would be even more so. [1] The find part is trickier. Discovering decent content is harder, as there's an arms race of ad-funded spammers trying to out-compete legitimate recipe sites in search-engine rankings. (There's also the possibility of search engines not being motivated to work on delivering good search results. [2]) The idea of having a choice between native GUI applications and web apps, has been with us for some time. Email is probably the best example, we've long had the choice between webmail and native email clients. Beyond webmail, these days even Microsoft Word has a web-based version. There are of course both advantages and disadvantages to web-based applications. [0] https://news.ycombinator.com/item?id=23730408 https://news.ycombinator.com/item?id=23730408 [1] https://news.ycombinator.com/item?id=29291392 https://news.ycombinator.com/item?id=29291392 [2] https://news.ycombinator.com/item?id=29772136 https://news.ycombinator.com/item?id=29772136

Karrot_Kream 5y ago
How do you know someone uses Gemini? They'll tell you the moment they can! Like the vegans of the web...
- antifa 5y ago
  I didn't want to register for a gemini group just to ask this, but I looked at it from a webdev point of view just long enough to notice there was no evidence of any way to implement a "save comment" or "type in search" feature, basically as if gemini was entirely reduced to mostly static websites with no forms of any kind possible. Is that an accurate assessment?
  - Karrot_Kream 5y ago
    > I didn't want to register for a gemini group just to ask this You actually don't need to. Gemini is a client-server architecture just like the web. You can grab a Gemini client, or use some of the web portals out there that make a server-side Gemini request and render the result to the browser. > or "type in search" feature There is a limited form of input allowed, but this input shows up as a query parameter essentially. This is how you can do search or offer some form of interactivity and how the current Gemini search engines/crawlers work. > basically as if gemini was entirely reduced to mostly static websites with no forms of any kind possible Again Gemini does allow limited forms, but it only accepts a single query parameter as input. You can, of course, parse the parameter however you choose, but the general culture is that there's very little interactivity and most everything is a static page.
jka 5y ago
Also entirely possible that I'm misunderstanding both of you, but I think what the parent comment was imagining was something like common schemas (like schema.org[1]?) for content that is currently around the web and encased in the challenging web design you mention. With common (and evolving) formats -- and incentives for publishers to provide their information within those formats -- we could then have much simpler, more streamlined tools to use and remix that data in application-specific ways. [1] - https://schema.org/docs/full.html https://schema.org/docs/full.html

hombre_fatal 5y ago

One problem is that it takes a lot of work and effort to build any of the valuable hubs where people post information. Ever try to start a forum? It's a monumental task with no guarantee of success. You may even need to employ people to grow and maintain one. And once you've finally grown one of these hubs that accumulates recipes, lyrics, real estate listings, classifieds, etc. (whatever you had in mind) there's no incentive to make it as easy as possible to share it with the world. Once you get over "ugh, everyone just wants to make a buck", there's the fact that it wasn't free to build and maintain the platform to begin with. And perhaps the only incentive to build the platform was the idea that people would pay for the value. Or, who is supposed to do the work of curating and organizing all of this information and then producing an API so that others can build on it, and why haven't they started? There are probably some inconvenient truths in the answer beyond cynicism.

anderspitman 5y ago
Ha, I'm trying to start a data ownership forum now. My approach has been to have it be a central place for support for all my open source projects. We'll see how that works out.
- antifa 5y ago
  > data ownership forum Care to share or explain more? Sounds awesome.
  - anderspitman 5y ago
    That's pretty much the whole tweet! Feel free to join us: https://forum.indiebits.io/ https://forum.indiebits.io/

dillondoyle 5y ago

Funnily, Google has pushed websites to add more structured data into their html for crawling.. Seems it's used for SEO hackinge. For instance on recipes why wouldn't a site give their recipe a super high rating. Those sites are awful SEO spam adservers basically. But business info that can be used to map seems pretty valuable to google. https://developers.google.com/search/docs/advanced/structured-data/search-gallery https://developers.google.com/search/docs/advanced/structure...

tweetle_beetle 5y ago
It's not as easy as faking metadata of 5* user ratings of your own content. If it ever was, that technique is no longer useful and likely to lead to losing rank. Google uses a subset of schema.org, validates it according to its own stricter specification and extends it in other ways. It doesn't naively consume everything you provide. For example [1] > If the Recipe structured data contains a single review, the reviewer's name must be a valid person or organization. For example, "50% off ingredients" is not a valid name for a reviewer. > Warning: If your site violates one or more of these guidelines, then Google may take manual action against it. Once you have remedied the problem, you can submit your site for reconsideration. [1] https://developers.google.com/search/docs/advanced/structured-data/review-snippet#guidelines https://developers.google.com/search/docs/advanced/structure...
- dillondoyle 5y ago
  Makes sense. And I'm sure they don't want to give too much away just helps SEO spammers even more

Mandatum 5y ago

I love RSS and MQTT for this as a "good enough" method. There's a few QL choices for CRUD operations too. We're getting there, albeit slowly..