4 ms·

If you’re open to sharing, what are you building?

adventured 7y ago
Sure. It's a new type of knowledge service (broadly in the Wikipedia, Quora, Britannica, Answers.com, wikiHow, Genius, et al. segment). Like most people I strongly dislike what has happened to the Web. I love the Web, I grew up on the Web, I'll never stop being in love with it (particularly what it once was, and still can be - it's a choice that we get to make in how we build). This service is Web only. It's built to be super fast, so it uses no bloated frameworks. It uses no trackers. It uses no traditional analytics services. It doesn't require piles of personal info and I don't want to know who you are. It has no social media (no social login or buttons or trackers or accounts). The typical large content page weighs in at around 20-40kb in size and loads in 150-200ms with 8 requests, no cookies, 7kb of script, 4kb of css (I can push that further, but it's tight enough for now). There will be no images on the entire service (SEO punishment inbound) other than a small text logo. No ads. No pop-ups. No dark patterns like hiding content. It doesn't require users / readers to use JavaScript. All content is Creative Commons licensed. The reason traditional VC can't be involved, is because they ultimately destroy all knowledge services they touch. I've been watching VCs (or private equity in the case of Answers.com; or giant corporations in the case of eg Verizon or IAC (eg about.com)) burn them to the ground for two decades now. To do it properly you have to prioritize what I call the knowledge motive, and absolutely minimize the profit motive. If you invert that value order, you'll destroy the service and you must inherently eventually betray the community that builds it up (this will happen to Stack Exchange; it has already happened to Quora). So this service is anti-commercialism; there is no business model going in; there will be no IPO; there is no exit plan; bare threads on expenses; it has to be run hyper thin, hyper optimized against cost and bloat. Doing that properly also helps keeps me independent. I have a ten year time frame allocated (including nearly two years into it so far), during which I'll push it forward and work on it no matter what happens, assuming I don't get hit by a bus or the equivalent. The ideal may or may not be to convert it to a charitable non-profit for long-lived purposes, assuming it works of course (I'm not assuming, I'm just doing it anyway because I want this thing to exist). I've messed with building various knowledge services going back a decade. This hit me like a bolt of lightning back in April of 2018 and I began to flesh it out. It required a few revisions in concept before full build phase began. I had to take about three months in there to teach myself some new things I had to know to be able to build this to operate very inexpensively from day one and forward (I'm assuming it will never be flush via operations). I'm probably going to do a Show HN this month or next for it and I'm going to begin showing it to some long-term Wikipedia editors that I know. Only a few people have seen it thus far, which breaks a good common rule - in this case I don't care, I'm not trying to find traditional product market fit, I don't have to desperately obey such concerns (it's solely my dime and my time), I built it because I wanted to, it will exist and persist because I want it to. There's a glorious freedom in that, and it might entirely fall flat initially or long-term - that's ok, I'm going in assuming the worst in terms of response; I'll see what kind of response it gets and decide on any adjustments. I'm in this for the very long-term and plotted that way from the beginning. It's another reason typical VCs can't be involved, their horizon is short-term oriented. They'll destroy you trying to force you to get big fast, and it won't matter to them if they do it, even if the thing could have succeeded by growing slower. It's just one of many reasons why knowledge services are the anti-VC segment. If you ever see a knowledge service take on big VC, you already know the guaranteed outcome ahead of time: burned to the ground, forced liquidation for pennies on the dollar, or an exit sale before the VC runs out to continue funding the bloat taken on during the get-big-too-fast phase (and then the buyer runs it into the ground or shuts it down, hello Freebase). So how will it stay afloat financially if it gains traction? Well, first, it's static cached text in the front, so it's dirt cheap to run at a very large scale. Beyond that I have some approaches that I think will work that I'll experiment with over time - it won't require much financially.
- Unsimplified 7y ago
  I'm interested in hearing more about its general design. I've been thinking about knowledge theory since 2012 and my current solution is to attach a broad privately-curated knowledge base (ended up simplifying to a markdown-git repo for now) on top of a high-tech engineering business for results-driven credibility/cashflow. Consider the modern landscape of information. Wikipedias are exhaustive and generic. Therefore inefficient (full of low-value ideas, distorted learning curve). Anti-promotion policy against spam also disrupts linking to great resources. Best for basic descriptions and factual details. Search engines are keyword-dependent and unfiltered. Therefore unfollowable (no idea progression map, ex. index) and requires active trust/relevance/value filtering (volatile, learning-rate-limited by current skill level). Best for targeted navigation (ex. find a local bus route) and first-step wide-index exploration (initial net thrown to catch better keywords/resources). Forums/Q&A involve waiting for answers within a permitted-QA context. Therefore slow, not guaranteed, indexed chronologically not logically, usually brief/incomplete. Best for popcorn-community exploration and immediate needs for new info.
  - adventured 7y ago
    > Forums/Q&A involve waiting for answers within a permitted-QA context. The question primes the brain for the answer. You can use that socratic lead structure without going with a traditional Q&A wait-and-pray approach (ala Stack Exchange, Answers.com, Yahoo Answers, and so on). It does require a specific system design though. You can induce and control the Q&As around a topic, you don't need to wait. Doing it intentionally and rigidly, in some cases leads to a far superior system, as opposed to the chaos and quality problems of junk filled Q&A sites like Yahoo Answers or Answers.com. In the history of the Web only one major traditional Q&A site has ever gotten it really right over time, that's Stack Exchange (and we'll see yet if their commercial interests don't erode what they accomplished, as the VCs demand their exit). The track record is abysmal because most Q&A sites are beholden to an inherently bad approach: relying heavily (and allowing) on large volumes of people with no specialization or passion for a subject to ask & answer questions (so you get a lot of low quality drive-by answers that have to be moderated away or tolerated). It's the equivalent of walking into McDonald's and expecting a five star experience, and then being surprised when it's not (when it was obvious all along exactly what was going to happen, only one outcome was possible). wikiHow, as one example, has persisted (while nearly all other how-to sites have died off in the age of Google Penguin/Panda/etc) at a modestly sound quality for so long, because they set the hows, rather than just waiting around for junk how-to questions to be asked and answered in a mediocre fashion by low quality drive-by contributors; and they accept a lower level of commercialism and volume. wikiHow worked because they do hows in a similar way conceptually to how Wikipedia does topic pages (it's all rather preordained down a strict funnel, instead of wild flailing). You can do the same thing in other ways in the Q&A segment. Not to mention of course in most cases a site's desperation for ad clicks and page views causes them to intentionally allow volumes of low quality trash to populate their Q&A systems (what Quora turned to as it became obvious they couldn't fulfill their valuation otherwise), instead of aggressively pursuing only quality. Quality in the knowledge space is very slow, it takes enormous amounts of time and requires aggressive, consistent, persistent moderation. It takes a long time to build a high quality knowledge culture that self-reinforces, self-protects. I don't personally believe all questions have merit, quite the opposite. This is another core flaw to the typical Q&A site. Few questions - in the grand scheme of all likely human-generated questions - have much wide merit. Numerous low value Q&A sites have overwhelmingly demonstrated that to be true over time. It's millions of people walking by, spitting on the sidewalk, and calling it art. > Therefore slow, not guaranteed, indexed chronologically not logically Slow is ok. Very few things of great value are built quickly, that's true today, and it has been true throughout history. Guaranteed you can heavily influence, by adding source requirements and restricting contribution (stepped barriers to contribution, and site-culture acting as an enforcer). I use a system that adds more friction to contribution than Wikipedia for example (I don't directly compete with them, I'm not building another encyclopedia), however there are aspects to my system that play to that approach better than it would on Wikipedia, so it evens out. You also want to build a culture that regulates low quality contributions, including brief / incomplete; you can do some of that technically, however you ultimately need a human culture involved at the center, I believe it'll still be another few decades at least before AI can do it effectively enough top to bottom. Indexing order can be influenced and dictated by editors into a logical structure, although this buckles under duress on the standard messy high volume Q&A sites with millions of people spitting on the sidewalk. Those are too disorganized, unstructured for that bottling / silo approach to work. The typical Q&A site is a landfill; landfills are mostly filled with high volumes of low value trash, you don't want to go in there and try to logically order it; it's a large amount of effort for a small payoff because the content isn't very valuable (does this rotten banana peel go before that one). You have to narrow the Q&As in topic, quality and volume, on the basis that not every question matters. If you believe every question & answer matters as a site, you end up as Yahoo Answers or Answers.com (ie worthless in the end).

Sure. It's a new type of knowledge service (broadly in the Wikipedia, Quora, Britannica, Answers.com, wikiHow, Genius, et al. segment). Like most people I strongly dislike what has happened to the Web. I love the Web, I grew up on the Web, I'll never stop being in love with it (particularly what it once was, and still can be - it's a choice that we get to make in how we build). This service is Web only. It's built to be super fast, so it uses no bloated frameworks. It uses no trackers. It uses no traditional analytics services. It doesn't require piles of personal info and I don't want to know who you are. It has no social media (no social login or buttons or trackers or accounts). The typical large content page weighs in at around 20-40kb in size and loads in 150-200ms with 8 requests, no cookies, 7kb of script, 4kb of css (I can push that further, but it's tight enough for now). There will be no images on the entire service (SEO punishment inbound) other than a small text logo. No ads. No pop-ups. No dark patterns like hiding content. It doesn't require users / readers to use JavaScript. All content is Creative Commons licensed. The reason traditional VC can't be involved, is because they ultimately destroy all knowledge services they touch. I've been watching VCs (or private equity in the case of Answers.com; or giant corporations in the case of eg Verizon or IAC (eg about.com)) burn them to the ground for two decades now. To do it properly you have to prioritize what I call the knowledge motive, and absolutely minimize the profit motive. If you invert that value order, you'll destroy the service and you must inherently eventually betray the community that builds it up (this will happen to Stack Exchange; it has already happened to Quora). So this service is anti-commercialism; there is no business model going in; there will be no IPO; there is no exit plan; bare threads on expenses; it has to be run hyper thin, hyper optimized against cost and bloat. Doing that properly also helps keeps me independent. I have a ten year time frame allocated (including nearly two years into it so far), during which I'll push it forward and work on it no matter what happens, assuming I don't get hit by a bus or the equivalent. The ideal may or may not be to convert it to a charitable non-profit for long-lived purposes, assuming it works of course (I'm not assuming, I'm just doing it anyway because I want this thing to exist). I've messed with building various knowledge services going back a decade. This hit me like a bolt of lightning back in April of 2018 and I began to flesh it out. It required a few revisions in concept before full build phase began. I had to take about three months in there to teach myself some new things I had to know to be able to build this to operate very inexpensively from day one and forward (I'm assuming it will never be flush via operations). I'm probably going to do a Show HN this month or next for it and I'm going to begin showing it to some long-term Wikipedia editors that I know. Only a few people have seen it thus far, which breaks a good common rule - in this case I don't care, I'm not trying to find traditional product market fit, I don't have to desperately obey such concerns (it's solely my dime and my time), I built it because I wanted to, it will exist and persist because I want it to. There's a glorious freedom in that, and it might entirely fall flat initially or long-term - that's ok, I'm going in assuming the worst in terms of response; I'll see what kind of response it gets and decide on any adjustments. I'm in this for the very long-term and plotted that way from the beginning. It's another reason typical VCs can't be involved, their horizon is short-term oriented. They'll destroy you trying to force you to get big fast, and it won't matter to them if they do it, even if the thing could have succeeded by growing slower. It's just one of many reasons why knowledge services are the anti-VC segment. If you ever see a knowledge service take on big VC, you already know the guaranteed outcome ahead of time: burned to the ground, forced liquidation for pennies on the dollar, or an exit sale before the VC runs out to continue funding the bloat taken on during the get-big-too-fast phase (and then the buyer runs it into the ground or shuts it down, hello Freebase). So how will it stay afloat financially if it gains traction? Well, first, it's static cached text in the front, so it's dirt cheap to run at a very large scale. Beyond that I have some approaches that I think will work that I'll experiment with over time - it won't require much financially.

Unsimplified 7y ago
I'm interested in hearing more about its general design. I've been thinking about knowledge theory since 2012 and my current solution is to attach a broad privately-curated knowledge base (ended up simplifying to a markdown-git repo for now) on top of a high-tech engineering business for results-driven credibility/cashflow. Consider the modern landscape of information. Wikipedias are exhaustive and generic. Therefore inefficient (full of low-value ideas, distorted learning curve). Anti-promotion policy against spam also disrupts linking to great resources. Best for basic descriptions and factual details. Search engines are keyword-dependent and unfiltered. Therefore unfollowable (no idea progression map, ex. index) and requires active trust/relevance/value filtering (volatile, learning-rate-limited by current skill level). Best for targeted navigation (ex. find a local bus route) and first-step wide-index exploration (initial net thrown to catch better keywords/resources). Forums/Q&A involve waiting for answers within a permitted-QA context. Therefore slow, not guaranteed, indexed chronologically not logically, usually brief/incomplete. Best for popcorn-community exploration and immediate needs for new info.
- adventured 7y ago
  > Forums/Q&A involve waiting for answers within a permitted-QA context. The question primes the brain for the answer. You can use that socratic lead structure without going with a traditional Q&A wait-and-pray approach (ala Stack Exchange, Answers.com, Yahoo Answers, and so on). It does require a specific system design though. You can induce and control the Q&As around a topic, you don't need to wait. Doing it intentionally and rigidly, in some cases leads to a far superior system, as opposed to the chaos and quality problems of junk filled Q&A sites like Yahoo Answers or Answers.com. In the history of the Web only one major traditional Q&A site has ever gotten it really right over time, that's Stack Exchange (and we'll see yet if their commercial interests don't erode what they accomplished, as the VCs demand their exit). The track record is abysmal because most Q&A sites are beholden to an inherently bad approach: relying heavily (and allowing) on large volumes of people with no specialization or passion for a subject to ask & answer questions (so you get a lot of low quality drive-by answers that have to be moderated away or tolerated). It's the equivalent of walking into McDonald's and expecting a five star experience, and then being surprised when it's not (when it was obvious all along exactly what was going to happen, only one outcome was possible). wikiHow, as one example, has persisted (while nearly all other how-to sites have died off in the age of Google Penguin/Panda/etc) at a modestly sound quality for so long, because they set the hows, rather than just waiting around for junk how-to questions to be asked and answered in a mediocre fashion by low quality drive-by contributors; and they accept a lower level of commercialism and volume. wikiHow worked because they do hows in a similar way conceptually to how Wikipedia does topic pages (it's all rather preordained down a strict funnel, instead of wild flailing). You can do the same thing in other ways in the Q&A segment. Not to mention of course in most cases a site's desperation for ad clicks and page views causes them to intentionally allow volumes of low quality trash to populate their Q&A systems (what Quora turned to as it became obvious they couldn't fulfill their valuation otherwise), instead of aggressively pursuing only quality. Quality in the knowledge space is very slow, it takes enormous amounts of time and requires aggressive, consistent, persistent moderation. It takes a long time to build a high quality knowledge culture that self-reinforces, self-protects. I don't personally believe all questions have merit, quite the opposite. This is another core flaw to the typical Q&A site. Few questions - in the grand scheme of all likely human-generated questions - have much wide merit. Numerous low value Q&A sites have overwhelmingly demonstrated that to be true over time. It's millions of people walking by, spitting on the sidewalk, and calling it art. > Therefore slow, not guaranteed, indexed chronologically not logically Slow is ok. Very few things of great value are built quickly, that's true today, and it has been true throughout history. Guaranteed you can heavily influence, by adding source requirements and restricting contribution (stepped barriers to contribution, and site-culture acting as an enforcer). I use a system that adds more friction to contribution than Wikipedia for example (I don't directly compete with them, I'm not building another encyclopedia), however there are aspects to my system that play to that approach better than it would on Wikipedia, so it evens out. You also want to build a culture that regulates low quality contributions, including brief / incomplete; you can do some of that technically, however you ultimately need a human culture involved at the center, I believe it'll still be another few decades at least before AI can do it effectively enough top to bottom. Indexing order can be influenced and dictated by editors into a logical structure, although this buckles under duress on the standard messy high volume Q&A sites with millions of people spitting on the sidewalk. Those are too disorganized, unstructured for that bottling / silo approach to work. The typical Q&A site is a landfill; landfills are mostly filled with high volumes of low value trash, you don't want to go in there and try to logically order it; it's a large amount of effort for a small payoff because the content isn't very valuable (does this rotten banana peel go before that one). You have to narrow the Q&As in topic, quality and volume, on the basis that not every question matters. If you believe every question & answer matters as a site, you end up as Yahoo Answers or Answers.com (ie worthless in the end).