19 votes

Topic deleted by author

32 comments

  1. [3]
    mrbig
    Link
    A fast, text-based website listing all the public cultural contests, grants, etc (for all the arts in Brazil) in a centralized and organized fashion, displaying essential information such as...

    A fast, text-based website listing all the public cultural contests, grants, etc (for all the arts in Brazil) in a centralized and organized fashion, displaying essential information such as deadlines, requirements, cash value, etc.

    6 votes
    1. [2]
      EgoEimi
      Link Parent
      Honestly, fast, clear, well-organized, text-based government databases and websites for everything would make modern civilization work 10x better. Screw making yet another CRM or food delivery...

      Honestly, fast, clear, well-organized, text-based government databases and websites for everything would make modern civilization work 10x better.

      Screw making yet another CRM or food delivery app. We should invest hundreds of billions of dollars in making good government websites with good UX for every entity at every level (in the USA where I am now).

      6 votes
      1. novov
        Link Parent
        Many governments are working towards this goal to some extent, with consistent design standards and technologies. For instance - although many of the individual sites are still outdated and crappy...

        Many governments are working towards this goal to some extent, with consistent design standards and technologies. For instance - although many of the individual sites are still outdated and crappy - we have https://www.govt.nz in my country. Of course, it is mired in all the usual bureaucracy and inefficiency, but the impetus is there.

        6 votes
  2. ruspaceni
    Link
    News globe website thing. Think google maps/earth but instead of it showing a bunch of shops and streetview, it would show news stories from different countries. It wouldn't even need to be fully...

    News globe website thing. Think google maps/earth but instead of it showing a bunch of shops and streetview, it would show news stories from different countries. It wouldn't even need to be fully featured as a directory/way to consume the content or anything, just a higher level view of the big events would suffice for a curious visitor.

    Time slider on the bottom (i looked into this part and made a 2d thing in mapbox a year or two back) so you can scrub through some more historic stuff, although that starts to become a lot of work with the scraping and whatnot.

    I've had this idea bouncing around in my head for 5-6 years at this point just silently hoping someone will beat me to it so i can just play around with it. But every now and then when i get a motivation burst I try to work on a different peice of the puzzle, which most recently was trying to see if I could write a parser to "pull out locations" from articles based on a huge dataset but all I could manage was a super slow false positive machine c: My next attempt at this is probably going to follow the basic idea of; go through text and identify proper nouns > google each noun and see what the internet says. So I'm keeping my eye out for api's and idly looking into the whole natural language programming side of things.

    Recently I've been thinking about ways to randomly generate a co-ordinate inside of specified country. Having 100s of stories in a single stack in the center of mass seems a bit counter intuitive for the 'broad overview' feel of the site, and if the randomized pins are a different colour then it would simply be inaccurate instead of misleading. Perhaps a traffic light system of sorts where it starts out roughly placed, and viewers can drag and drop them to form a more accurate consensus (but again that becomes a lot of work, very fast)

    I'm convinced that it'd be a cute little website but also a very disproportionate amount of work for the somewhat specific version I have in my head. I'm happy even if it never comes to fruition though because it's been a good source of personal challenges where there's no proper ways to do it, so it doesn't matter how hacky my implementation is as long as I'm learning and making progress.

    4 votes
  3. [2]
    Bullmaestro
    Link
    FYI: I have next to no web dev experience. I can barely even program a basic webpage using PHP and MySQL. Constructing any of the below ideas are well below my skillset. Few ideas I had bouncing...

    FYI: I have next to no web dev experience. I can barely even program a basic webpage using PHP and MySQL. Constructing any of the below ideas are well below my skillset.

    Few ideas I had bouncing in my head lately.

    1) A decentralised social network similar to Reddit, Facebook, etc that uses technologies like WebTorrent to save on hosting costs/infrastructure. Not only is a social network pretty challenging to make as-is, but I can imagine this being a nightmare to moderate or police.

    2) A website that allows writers and aspiring authors to upload, publish and share their work. Sites like Fanfiction.net and Wattpad already exist, but this would be a bit different. It would have a partner program which would offer advertisement revenue sharing with popular users, plus the ability for premium subscribers to download any stories on the site and read offline via PDF, EPUB or other e-reader formats, or even allow the sale of paperback/hardback versions of work on a print-on-demand basis.

    I can see this being very challenging from a copyright perspective due to how much work it would take to purge infringing content off the site.

    3) User written news site with revenue-sharing, similar to Newsvine. Newsvine was shut down four years ago by MSNBC. I think the problem that eventually led to its downfall was the fact that it was bought out by a big news network.

    4 votes
    1. mxuribe
      Link Parent
      As i began to read your point here, I was going to reply with "the fediverse" already exists (accessible via mastodon, pleroma, etc.)...but then upon reading the latter portion understood that...
      1. A decentralised social network similar to Reddit, Facebook, etc that uses technologies like WebTorrent to save on hosting costs/infrastructure.

      As i began to read your point here, I was going to reply with "the fediverse" already exists (accessible via mastodon, pleroma, etc.)...but then upon reading the latter portion understood that (while decentralized/distributed) the fediverse isn't built on tech like torrents nor blockchains, etc....but then wondered if maybe Secure Scuttlebutt (SSB) might be more along the lines of what you were thinking? See: https://en.wikipedia.org/wiki/Secure_Scuttlebutt

  4. lionirdeadman
    Link
    I've wanted to make a website which aggregates RSS feeds from personal websites and you can essentially just full-text search through it or filter by tag. It doesn't sound very hard to make. This...

    I've wanted to make a website which aggregates RSS feeds from personal websites and you can essentially just full-text search through it or filter by tag. It doesn't sound very hard to make.

    This is inspired by this post.

    3 votes
  5. guts
    Link
    A decentralized TikTok.

    A decentralized TikTok.

    3 votes
  6. [11]
    tomf
    Link
    it would assign a universal ID for every book, not every edition like ISBN... and it would have series info, plots, language selection, an open API, etc. :)

    it would assign a universal ID for every book, not every edition like ISBN... and it would have series info, plots, language selection, an open API, etc. :)

    2 votes
    1. [10]
      petrichor
      Link Parent
      Archive.org's OpenLibrary seems close to what you're looking for.

      Archive.org's OpenLibrary seems close to what you're looking for.

      2 votes
      1. [9]
        tomf
        Link Parent
        I wish! haha. Check this vs goodreads vs Google Books API. Right now I'm pulling RSS from Goodreads > IFTTT > Google Sheets for the title, series, original date, and pages -- then I use a script...

        I wish! haha. Check this vs goodreads vs Google Books API.

        Right now I'm pulling RSS from Goodreads > IFTTT > Google Sheets for the title, series, original date, and pages -- then I use a script to get a good Google Books ISBN to pull the main endpoint like above to fill in the plot.

        LOC and other sources are all good if you have IDs, but most still don't bother with the series info, which is really important. I've been bitching about ISBNs and the lack of a central ID for titles for years now. If I ever meet someone who sympathizes with the struggle, it'll be wedding bells.

        4 votes
        1. [4]
          GoingMerry
          Link Parent
          I’m working on a book website myself but haven’t run into the universal ID issue yet. Just letting you know so you can save up for my engagement ring.

          I’m working on a book website myself but haven’t run into the universal ID issue yet. Just letting you know so you can save up for my engagement ring.

          4 votes
          1. [3]
            tomf
            Link Parent
            I hope you like caviar. :) The main beefs I have with the book info is that sometimes it'll return a clean plot, where other times it'll simply return the back cover (e.g. 'FROM THE #1 BESTSELLING...

            I hope you like caviar. :)

            The main beefs I have with the book info is that sometimes it'll return a clean plot, where other times it'll simply return the back cover (e.g. 'FROM THE #1 BESTSELLING AUTHOR OF THE HARRY BOSCH AND LINCOLN LAWYER SERIES') and most sites don't include the full series info (e.g. Mickey Haller #1; Harry Bosch Universe #15 for the Lincoln Lawyer), and also defaulting to the original published date. I don't care if a book was reprinted in 2014, I want 1851 for Moby Dick.

            2 votes
            1. [2]
              GoingMerry
              Link Parent
              I’ve been doing interviews in the industry for my project and the data segmentation and cleanliness issues are pretty much due to publishers setting up walled gardens. They are extremely powerful....

              I’ve been doing interviews in the industry for my project and the data segmentation and cleanliness issues are pretty much due to publishers setting up walled gardens. They are extremely powerful.

              I’d love to usurp them via technology but it’s a tough nut to crack!

              2 votes
              1. tomf
                Link Parent
                it's really bizarre how late to the game they are with... everything. A centralized, standardized reference for books would benefit everybody involved. I finally have my little spreadsheet...

                it's really bizarre how late to the game they are with... everything. A centralized, standardized reference for books would benefit everybody involved.

                I finally have my little spreadsheet finished. While it isn't perfect, its good enough for the task at hand. I can only imagine the longsuffering involved in cleaning a dataset of all books.

        2. [4]
          krg
          Link Parent
          What do you mean by “series info”? Also, would you have separate IDs for different revisions of the same book? In your scheme, these IDs wouldn’t represent a physical entity, correct? So, a...

          What do you mean by “series info”?

          Also, would you have separate IDs for different revisions of the same book? In your scheme, these IDs wouldn’t represent a physical entity, correct? So, a collection that has many “books” in a volume would have its own ID that links to the IDs of the books within?

          1 vote
          1. [3]
            tomf
            Link Parent
            Michael Connelly books are a good example of a series. We have individual series and the overall world series. Title Year Series The Black Echo 1992 Harry Bosch #1; Harry Bosch Universe #1 The...

            Michael Connelly books are a good example of a series. We have individual series and the overall world series.

            Title Year Series
            The Black Echo 1992 Harry Bosch #1; Harry Bosch Universe #1
            The Black Ice 1993 Harry Bosch #2; Harry Bosch Universe #2
            Blood Work 1998 Terry McCaleb #1, Harry Bosch Universe #8
            Chasing the Dime 2002 Henry Pierce #1; Harry Bosch Universe #10.5
            The Lincoln Lawyer 2005 Mickey Haller #1; Harry Bosch Universe #15
            The Brass Verdict 2008 Harry Bosch #14; Mickey Haller #2; Harry Bosch Universe #18
            The Scarecrow 2009 Jack McEvoy #2; Harry Bosch Universe #19
            Blue on Black 2010 Harry Bosch #14.5; Harry Bosch Universe #20.5
            The Late Show 2017 Renée Ballard #1; Harry Bosch Universe #29
            Two Kinds of Truth 2017 Harry Bosch #20; Harry Bosch Universe #30
            Dark Sacred Night 2018 Renée Ballard #2; Harry Bosch #21; Harry Bosch Universe #31
            The Night Fire 2019 Harry Bosch #22; Renée Ballard #3; Harry Bosch Universe #32
            Fair Warning 2020 Jack McEvoy #3; Harry Bosch Universe #33
            The Law of Innocence 2020 Mickey Haller #7; Harry Bosch Universe #34

            or

            Title Year Series
            A Drink Before the War 1994 Kenzie & Gennaro #1
            Darkness, Take My Hand 1996 Kenzie & Gennaro #2
            Sacred 1997 Kenzie & Gennaro #3
            Gone, Baby, Gone 1998 Kenzie & Gennaro #4
            Prayers for Rain 1999 Kenzie & Gennaro #5
            Moonlight Mile 2010 Kenzie & Gennaro #6

            Ideally we'd treat it like we treat movies with IMDB IDs but with other releases listed as sub-IDs. If you want the basic info on Kill Bill Vol1 and Vol2 you can pull that up, but if you want to pull up the info regarding the Japanese DVDs that have some extra scenes and one fight scene in color, you could get that too, sort of like a tv series.

            Goodreads has the best info out of them all, but their API is pretty much closed off. They have a mix of XML and JSON and often you'll pull a dozen books and get foreign language info returned, etc etc.

            2 votes
            1. [2]
              Adys
              Link Parent
              What about wikidata, what do they have/use?

              What about wikidata, what do they have/use?

              2 votes
              1. tomf
                Link Parent
                I couldn't ever get it to give me a clean output. This is a sample of the page I usually end up with -- https://www.wikidata.org/wiki/Q1081911 It does give links to the wikipedia page if there is...

                I couldn't ever get it to give me a clean output. This is a sample of the page I usually end up with -- https://www.wikidata.org/wiki/Q1081911

                It does give links to the wikipedia page if there is one, but that doesn't have series. It wouldn't be the worst for descriptions, but that still relies on each book having its own page, which isn't always the case.

                It does differentiate between work and edition, which is nice, though.

                2 votes
  7. [2]
    vektor
    Link
    Try out distributed trust, i.e. a website for, say reviews, where you can mark reviews/users you agree with/trust. ML/Stats behind the scenes works out which reviews you should trust, kind of like...

    Try out distributed trust, i.e. a website for, say reviews, where you can mark reviews/users you agree with/trust. ML/Stats behind the scenes works out which reviews you should trust, kind of like collaborative filtering.

    Say goodbye to (a) review sites run by orgs with a conflict of interest (sellers, e.g.) and say goodbye to sellers shilling on independent sites.

    2 votes
    1. [2]
      Comment deleted by author
      Link Parent
      1. vektor
        Link Parent
        Oh, basically, the idea of collaborative filtering (a recommendation system design) is that you want to recommend the content that is watched by similar people. So find someone else on your...

        Oh, basically, the idea of collaborative filtering (a recommendation system design) is that you want to recommend the content that is watched by similar people. So find someone else on your platform that watches what I watch, recommend to me what he watches, boom, done. Makes the numbers reasonably easy to crunch, as you don't even have to analyze the content.

        Ok, but that algorithm can be used for things other than recommendations. We can also expose it to the user. Say you're looking at a bunch of reviews for a product, and the system computes the chance that you'd agree with any given review, based solely on who wrote it. This author has shilled a few products before that you did not like, or maybe he has highly rated a product that people you trust found horrible. Maybe you shouldn't give a damn what that reviewer thinks.

        This other reviewer has submitted a more critical review, and their other reviews generally line up with your or those you trust. Maybe you should trust that guy more. Ignore all the untrustworthy reviews, and this 4.5 star rated off-brand affordable product is actually a cheap knock-off at 3 stars and not worth your time. There's a few things to be solved here on the ML side of things, e.g. trust-by-degrees-of-separation and the inherent noisiness of customer satisfaction - if you liked a product but someone else had a defective unit, does that make their reviews untrustworthy? Of course not, but the algorithm can't tell.

        Does that make it more clear?

        Oh, and I've got another one: A e-commerce site where customers dictate the terms: Say for example I would want to switch my power company (using this one, because due to interchangeability of service it's low hanging fruit), but I have terms I would be unwilling to accept. I could specify these terms and indicate the price I would be willing to pay. Companies can browse the site looking for customers. The idea is to give customers a way to amass their collective power: One guy demanding that their power company allow them to cancel monthly isn't going to change anything. If a power company can see that they could have 1000 new customers if they change their terms a bit, maybe we can get some change going. Another market would be appliances, where customers could demand more favorable warranty terms: It is entirely possible for manufacturers to build them to last quite long, it just isn't economical. If a lot of people would be willing to buy e.g. washing machines with 15 year warranty, the manufacturer could justify building them to last at least that long.

        2 votes
  8. [8]
    Merry
    Link
    A site like Tildes/Reddit, but instead of writing out a comment you speak it. Would be a slower form of Clubhouse.

    A site like Tildes/Reddit, but instead of writing out a comment you speak it. Would be a slower form of Clubhouse.

    1 vote
    1. stu2b50
      Link Parent
      Honestly that does sound interesting. The issues I imagine are that audio files are going to be much more expensive to host, and that the high barrier of entry could limit user base growth even...

      Honestly that does sound interesting. The issues I imagine are that audio files are going to be much more expensive to host, and that the high barrier of entry could limit user base growth even more than it does now.

      I could try and whip up a prototype when I have some time.

      3 votes
    2. [2]
      Bullmaestro
      Link Parent
      I worry it'll turn into something like this very fast.
      1 vote
      1. stu2b50
        Link Parent
        That's the reverse, though - type text and have tts speak it. Typing text is low cost (especially when you can copy paste things), listening to it is much higher cost. For this hypothetical...

        That's the reverse, though - type text and have tts speak it. Typing text is low cost (especially when you can copy paste things), listening to it is much higher cost. For this hypothetical inputting becomes the bottleneck.

        2 votes
    3. [5]
      Comment deleted by author
      Link Parent
      1. [4]
        Merry
        Link Parent
        https://record.reverb.chat/s/1bB5Yy0LhhvUXpMtPjcj
        1 vote
        1. vektor
          Link Parent
          Actually heard of a dating app that uses this premise. Basically, your profile is images and voice messages; if you match with someone you then send voice messages back and forth.

          Actually heard of a dating app that uses this premise. Basically, your profile is images and voice messages; if you match with someone you then send voice messages back and forth.

          2 votes
        2. Neuroflux
          Link Parent
          Lacking a website dedicated to this purpose, you could base a subreddit or group on Tildes around the concept. On reddit, you could have automod automatically delete any comment/reply that isn't a...

          Lacking a website dedicated to this purpose, you could base a subreddit or group on Tildes around the concept. On reddit, you could have automod automatically delete any comment/reply that isn't a link to an audio recording.

          1 vote
        3. MeckiSpaghetti
          Link Parent
          Yeah! I think people would be so much nicer to each other if everything was spoken word 😊

          Yeah!

          I think people would be so much nicer to each other if everything was spoken word 😊

          1 vote
  9. Kuromantis
    Link
    Is there a specific aspect of website building you want to be challenged on? The 2 ideas I would try if I could program (and had all the free time in the world lol) are: 1: A website where you...

    Is there a specific aspect of website building you want to be challenged on? The 2 ideas I would try if I could program (and had all the free time in the world lol) are:

    1: A website where you write some text and the site tells you for how long that word has had (one of?) it's current meaning or existed. I wouldn't choose to display this via multiple <details> looking clickable areas with the time period labeled on them where the words highlighted would not be the same in that time period, so do that so it's "not offered the way that you want".

    2: You basically look for historical footage of (in my mind political but realistically there's lots of sports footage where people do similar things) speeches where the audience at some point cheers or claps in unison and you basically try to sort them (in whatever way you prefer), let people hear random ones or automatically generate ones based on what you already have.

    However, a lot of the challenge here would be finding the data for words and speeches, not the websites, so I'm not sure if these are viable suggestions.

    1 vote
  10. moocow1452
    Link
    A website similar to Justwatch.com but for comic books and manga where you can make a pull list, and see what services have what books available.

    A website similar to Justwatch.com but for comic books and manga where you can make a pull list, and see what services have what books available.

    1 vote
  11. PetitPrince
    Link
    A Start to Crate database, with perhaps different way to prove the StC time (screenshot, video, ... basically it's a niche speedrun category).

    A Start to Crate database, with perhaps different way to prove the StC time (screenshot, video, ... basically it's a niche speedrun category).

    1 vote