Ask HN: What's broken about the internet?
46 comments
These days you need an internet-device and an anti-internet-device like a Pi-Hole plus half a dozen browser plugins. Every driver of every part of your computer is trying to sneak out some of you data. Every website is full of trackers. You are not only fighting off criminals but also some of the most powerful corporations and every government. A hand full of companies run large parts of the infrastructre - you couldn't live without Google, Facebook, Microsoft, Apple and Amazon if you wanted. The central social networks are hate and manipulation machines. Everything wants to sell you something. And when you pay to buy it, it tries to steal your data anyways.
This is what makes me the saddest. The internet has so much potential for good, but it doesn't appear to be heading that direction.
Every couple months I get an urge to give up on tech and start subsistence farming in the middle of nowhere.
Every couple months I get an urge to give up on tech and start subsistence farming in the middle of nowhere.
I was doing some work for a company and someone posted on Slack about ad blockers. I noted to everyone that I used ad blockers because I don't like how invasive ads have become.
The founder and CTO then proceeded to berate me in the channel for using a blocker. He had made a lot of money over the years from writing code that helped ads figure out what content to target for a user.
I still use an ad blocker, but have added Disable Javascript to all my browsers as well, just in case they try to detect I'm blocking things.
The founder and CTO then proceeded to berate me in the channel for using a blocker. He had made a lot of money over the years from writing code that helped ads figure out what content to target for a user.
I still use an ad blocker, but have added Disable Javascript to all my browsers as well, just in case they try to detect I'm blocking things.
In no particular order:
- The WWW - from a delightfully whimsical follow the rabbit-hole experience in the early 90s where discovery was hard, but content was king; to a centralised, AI-ranked, viral echo-chamber in the 00s and 10s where the discovery engines now control the whole experience and the content you see is presented based on who pays. The key problem here being the monetisation of discovery. We need solve search and discovery, but deliver it as an open Internet standard like DNS or HTTP.
- Smart devices - buying an appliance with a closed-source, embedded device that relies on Internet connectivity and the solvency of it's manufacturer in order to operate it, patch it, secure it, and maintain it is the antithesis of what this planet needs right now. When the CA Root certificate(s) installed in your no-name smart TV expires and the OEM doesn't exist/doesn't care to provide you a firmware update, these devices will become less than worthless and most likely landfill. We need industry to adopt an open framework for smart devices that helps prolong their lifespan eg: a public Linux repo for updates to the underlying OS - the OEMs can deploy their own user interface, but end-users should be able to pick and choose if they wish (and most wont').
- Trust - in particular X.509 certificates. A lot of progress has been made in making trust via digital certificates the default rather than a paranoid exception, with a large portion of the web being delivered over HTTPS, RPKI for BGP currently being deployed in large operators and DNSSEC showing some (admittedly slow) signs of adoption. What is still a major problem in this area is the complexity of certificate management and renewal. The work LetsEncrypt and the EFF (certbot) have done in automating this process is fantastic, but these are still a long way from mainstream usage.
- Smart devices - buying an appliance with a closed-source, embedded device that relies on Internet connectivity and the solvency of it's manufacturer in order to operate it, patch it, secure it, and maintain it is the antithesis of what this planet needs right now. When the CA Root certificate(s) installed in your no-name smart TV expires and the OEM doesn't exist/doesn't care to provide you a firmware update, these devices will become less than worthless and most likely landfill. We need industry to adopt an open framework for smart devices that helps prolong their lifespan eg: a public Linux repo for updates to the underlying OS - the OEMs can deploy their own user interface, but end-users should be able to pick and choose if they wish (and most wont').
- Trust - in particular X.509 certificates. A lot of progress has been made in making trust via digital certificates the default rather than a paranoid exception, with a large portion of the web being delivered over HTTPS, RPKI for BGP currently being deployed in large operators and DNSSEC showing some (admittedly slow) signs of adoption. What is still a major problem in this area is the complexity of certificate management and renewal. The work LetsEncrypt and the EFF (certbot) have done in automating this process is fantastic, but these are still a long way from mainstream usage.
Well, the biggest problems with the internet seems to be the World Wide Web. NNTP has improved (some of these improvements at my suggestions, such as support for 63-bit article numbers), and Gopher has also improved since the original specification (now there are "i" type lines in menus, which are useful), but WWW has just gotten worse (and that includes Hacker News to some degree too, but not as much as Google and Facebook and so on). I have set up not only HTTP but also NNTP, Gopher, QOTD, SMTP, etc, and may later also set up IRC, Telnet, Viewdata, etc. Additionally, many things I serve on the HTTP are just direct downloads anyways; no need to deal with 100 megabyte files just to access a text file or other file that you wanted to download. And, you can do it too, if you want to do, I suppose.
Content mills -
Someone wise said that tools that make writing easier turn bad writers into worse writers.
Goodhart's law means that nearly all content is broken. It is judged by its view count and Google rank, so all it is good for is getting clicks and ranking well.
So much effort, so much money, ploughed into creating really really awful content which hides away the 10% that isn't crap (Sturgeon?).
Someone wise said that tools that make writing easier turn bad writers into worse writers.
Goodhart's law means that nearly all content is broken. It is judged by its view count and Google rank, so all it is good for is getting clicks and ranking well.
So much effort, so much money, ploughed into creating really really awful content which hides away the 10% that isn't crap (Sturgeon?).
Also targeted advertising
Business model seems to be broken. I pay a giant internet bill each month, basically insane margins based on what it should actually cost to deliver much faster connectivity than what I get, yet none of that extra money funds the things that I actually use the internet for. (Why does it cost that much? Part of the reason may be that our local cable company is a government-enforced monopoly, that it's next to impossible to get right of way for new fiber, and that wireless is currently uncompetitive. Perhaps 5G may change things, but I'm not holding my breath.)
Moreover, the internet makes distribution nearly free and allows a nearly unlimited number of people to access information and digital media from all over the world, but lengthy (70 years or more) copyright terms make it illegal to do so in many cases. Instead, thousands or millions of person-hours are spent on the impossible task of trying to make bits behave like physical objects in order to satisfy legal and business requirements. When an organization such as the internet archive tries to make a digital library whose collection isn't bound by the constraints of physical libraries, they are sued by publishers for copyright infringement and potentially liable for $150k in statutory damages per occurrence.
Moreover, the internet makes distribution nearly free and allows a nearly unlimited number of people to access information and digital media from all over the world, but lengthy (70 years or more) copyright terms make it illegal to do so in many cases. Instead, thousands or millions of person-hours are spent on the impossible task of trying to make bits behave like physical objects in order to satisfy legal and business requirements. When an organization such as the internet archive tries to make a digital library whose collection isn't bound by the constraints of physical libraries, they are sued by publishers for copyright infringement and potentially liable for $150k in statutory damages per occurrence.
IANAE, but it seems like the "The web is for transfer of information" and "The web is for applications" ideas should be split into two separate domains. I shouldn't need an OS in order to read HN or browse a text based site.
Yeah, I'm thinking more and more the same, I'm beginning to think we need a dual function web.
Static web, and run as an app web.
I kinda have this by running with javascript off (this solves all sorts of other issues, though much of the web doesn't work that could work.) When I need to 'run as app' I allow js for that one visit.
Static web, and run as an app web.
I kinda have this by running with javascript off (this solves all sorts of other issues, though much of the web doesn't work that could work.) When I need to 'run as app' I allow js for that one visit.
That's what I do on most machines I use, but it just feels kinda wrong to not only have to use a bloated app, but also install add-ons to said app, when I just want to read some text
This is probably just me being picky though.
This is probably just me being picky though.
Smart comments. I don't think you're being picky. I'm kinda shocked that this is the only thread about reading because that's what I think is wrong with the web: It's not good for reading. I'm 4 years into co-founding a startup that's solving this problem.
Making the internet more useful in the area of concise to the point information without sidetracks would be a big improvement. How is your startup approaching the reading problem on the web?
Privacy and removal of insidious tracking would be a good start.
I feel the web is too centralised, with half a dozen or so platforms essentially being gatekeepers of content on the web.
People link out on their sites much less tha in the past, in the belief that it raises the chance of penalising them on Google and hurt their rankings. Last I looked search engine are typically responsible for delivering around 50% of visitors to a site, and Google has a near monopoly in many countries.
Wikipedia while great provides a less than obvious set of rules and regulations before adding data into it. It typically tends to rank first on all major search engines for any query.
Social media have become moral compasses in what is OK and what is not OK to talk about.
A more diversified web moving away from these 'decision makers' IMO would make it a healthier place.
I feel the web is too centralised, with half a dozen or so platforms essentially being gatekeepers of content on the web.
People link out on their sites much less tha in the past, in the belief that it raises the chance of penalising them on Google and hurt their rankings. Last I looked search engine are typically responsible for delivering around 50% of visitors to a site, and Google has a near monopoly in many countries.
Wikipedia while great provides a less than obvious set of rules and regulations before adding data into it. It typically tends to rank first on all major search engines for any query.
Social media have become moral compasses in what is OK and what is not OK to talk about.
A more diversified web moving away from these 'decision makers' IMO would make it a healthier place.
Make it more decentralized ... why should stupid laws made in the US impact my experience ???.
The lack of privacy and security. Throwing an ad-blocker on a browser is just a band-aid; the festering wound is still there getting worse.
My world view (or more like a cosmic view) is this: Space and matter are fascinating phenomenon but the truly fascinating things happen between the two. Empty space and time are quite boring until you put matter in it.
The "Let the output of your program be the input of mine." philosophy was really good. Web applications do the exact opposite. All of the interesting shit is missing.
Keeping parts of a thing in separate places can be necessary at times but it usually is not. There is the layout of an article website and there are article in it. The articles with their videos and images are separate things as much as browsers are separate things from web pages (perhaps even more so).
The article can be a separate file. Like an iframe but as inline text. No <head>, no <script>, no <style> but if it has any of those they are ignored.
The other perspective also works. The data for a physical object could be a single file. One or more images, videos, docs, the meta data. Say a product, its price and description can be put inside the image file and be made accessible as a js object/json. Standards can be created, adopted and popularized. If the picture has one or more cars the meta data can have all the usual properties of a car with standardized key names. This seems chaotic and undesirable for large shopping platforms but if an object has a description, contact information and a price it can be sold. Setting up a small store could be as simple as dropping images into a page. One would want to parse out some of the meta data of course but there can be much more data available.
It makes reasoning about the website much more like we reason about the real world.
The "Let the output of your program be the input of mine." philosophy was really good. Web applications do the exact opposite. All of the interesting shit is missing.
Keeping parts of a thing in separate places can be necessary at times but it usually is not. There is the layout of an article website and there are article in it. The articles with their videos and images are separate things as much as browsers are separate things from web pages (perhaps even more so).
The article can be a separate file. Like an iframe but as inline text. No <head>, no <script>, no <style> but if it has any of those they are ignored.
The other perspective also works. The data for a physical object could be a single file. One or more images, videos, docs, the meta data. Say a product, its price and description can be put inside the image file and be made accessible as a js object/json. Standards can be created, adopted and popularized. If the picture has one or more cars the meta data can have all the usual properties of a car with standardized key names. This seems chaotic and undesirable for large shopping platforms but if an object has a description, contact information and a price it can be sold. Setting up a small store could be as simple as dropping images into a page. One would want to parse out some of the meta data of course but there can be much more data available.
It makes reasoning about the website much more like we reason about the real world.
Trolls, spam and ads. IoT - Internet of turds. GAFAM taking away control of our devices in pursuit of surveillance economy. But generally technical illiteracy of average consumer is getting worse.
Certain web browser owned by the largest advertising company is playing with the idea of deprecating URLs. Maybe Mozilla Builders can find a way to deprecate TCP/IP so that we can start from scratch until eternal september mark ii hits us.
Server farms are weird concept to me. I feel the infrastructure the internet is build upon is a bit archaic and could use a more rewrite. not sure how though
Search. Google results are now full of SEO optimized low quality affiliate garbage.
I recently tried switching my default search engine to duckduckgo for like the 4th time in my life (I've been trying it about once a year), and this time I'm sticking with it. It's good enough now that I don't need to use google for every search. Maybe you could try again.
A search engine can be much better than google. All shopping websites of any significance have better features. (Is google treating links as competing technology?)
Denying the search provider a page view might work. Have them return xml. Use the OS, browser or custom theme. Build a search extension platform on top of it, sit back and watch what people come up with. (I would filter out websites that need more than x requests besides images and video, have more than y kb js, z kb css, make big news websites into a check box in the side menu, I can think of 30 extensions right now)
I made a search bar for Opera one time (widget) that had a row of submit buttons (icons) with different targets. It sat on top so one could just click the next button. It is hard to explain but it can be quite surprising to re-send your query some place else.
MyIE use to be able to open X search results in background tabs. A mouse gesture would close one and move to the next.
Denying the search provider a page view might work. Have them return xml. Use the OS, browser or custom theme. Build a search extension platform on top of it, sit back and watch what people come up with. (I would filter out websites that need more than x requests besides images and video, have more than y kb js, z kb css, make big news websites into a check box in the side menu, I can think of 30 extensions right now)
I made a search bar for Opera one time (widget) that had a row of submit buttons (icons) with different targets. It sat on top so one could just click the next button. It is hard to explain but it can be quite surprising to re-send your query some place else.
MyIE use to be able to open X search results in background tabs. A mouse gesture would close one and move to the next.
Search is shit because of advertising influence on algorythm design and tuning.
Having observed that major search engines often seem to hide what I'm actually looking for - web search results - "below the fold" (assuming it's there at all) I can't help but wonder if there is an opportunity for building niche "web search" engines that search the web (rather than the adverWeb, the seoWeb, or the goog/twit/faceWeb.)
As I've mentioned before, my favorite flavor of Google is Google Scholar. I'd also like to be able to slice/subset search results to get scientific, non-commercial, public interest, etc..
As I've mentioned before, my favorite flavor of Google is Google Scholar. I'd also like to be able to slice/subset search results to get scientific, non-commercial, public interest, etc..
discovery . basically discovery through google is bimodal: either the same 5 top sites, or horrible SEO trash. Gimme answers from forums when i m looking for them, please
Mobile-everything craze. No we don't need to turn everything to information-sparse, inflexible mobile replicas. Give us more density, there's still a zoom-in function
kill the cloud. re-decentralize
Mobile-everything craze. No we don't need to turn everything to information-sparse, inflexible mobile replicas. Give us more density, there's still a zoom-in function
kill the cloud. re-decentralize
UI toolkits - Seems that all the highest quality UI toolkits (Ant, Material, Fabric/Fluent, etc) have way too much whitespace. It works for things that work well in "cards" but the low information-density is killing me personally. I think more developers would choose high-density UI toolkits if one was released by a major dev team and could be relied on to be maintained. Maybe Photon/Protocol could eventually fill this gap, I like them a lot more overall but they don't seem nearly as well-packaged for easy use, or at least don't seem to have a complete kit available at https://www.adobe.com/products/xd/resources.html or anywhere else really.
Walled gardens - not being able to build things that integrate multiple services. Instagram live videos, combining amazon and walmart listings, google places "here now", or unified client for iOS/SMS/Discord/Slack/Teams are all off limits to ambitious developers.
Centralization - Seems like home email servers are increasingly being ban-hammered by the major providers like Gmail, ProofPoint, etc so you have to have your mail sent by a hosted server. CloudFlare is also seemingly becoming a single point of denial/failure. Chromium can now effectively dictate W3C standards because whatever they implement instantly becomes the ground truth for web dev.
DNS/BGP - Sorely in need of a major update.
IoT - It's shockingly difficult to find IoT devices which can compute locally/privately/on-the-edge. Everything sends all the data back home and stores/processes it there.
Tracking - If websites insist on having an OAuth or email verified signup, I'd like my browser to quickly and quietly create a new profile for every domain, or even every visit to that domain. And otherwise keep all tracking/fingerprinting down to effectively zero. I'm really tired of being tracked everywhere all the time.
Walled gardens - not being able to build things that integrate multiple services. Instagram live videos, combining amazon and walmart listings, google places "here now", or unified client for iOS/SMS/Discord/Slack/Teams are all off limits to ambitious developers.
Centralization - Seems like home email servers are increasingly being ban-hammered by the major providers like Gmail, ProofPoint, etc so you have to have your mail sent by a hosted server. CloudFlare is also seemingly becoming a single point of denial/failure. Chromium can now effectively dictate W3C standards because whatever they implement instantly becomes the ground truth for web dev.
DNS/BGP - Sorely in need of a major update.
IoT - It's shockingly difficult to find IoT devices which can compute locally/privately/on-the-edge. Everything sends all the data back home and stores/processes it there.
Tracking - If websites insist on having an OAuth or email verified signup, I'd like my browser to quickly and quietly create a new profile for every domain, or even every visit to that domain. And otherwise keep all tracking/fingerprinting down to effectively zero. I'm really tired of being tracked everywhere all the time.
Search because web spam.
I've been trying to get a blackhat webspam site down (it has child porn and piracy on it, besides spam links). But no luck so far, they effectively fool Googlebot and ruin search rankings.
I've been trying to get a blackhat webspam site down (it has child porn and piracy on it, besides spam links). But no luck so far, they effectively fool Googlebot and ruin search rankings.
Talking about the www specifically: too many trackers, dark patterns, hard to use sites, cookie notifications are just an annoyance and often come with dark patterns and no opt outs, ridiculous asset bundle sizes, auto play videos, pop ups, ridiculously large attack surface called the web api, working group standards ==
Chrome is the standard, Chinese firewall, browser complexity makes it hard to compete, JS security challenges ...
The social media walled garden - quite a few people I know rarely use a browser or don't even know how to use it. They get their news mostly from their social media apps, and they have a hard time spotting fake news. It is much easier to censor content in a social network than the open internet. People live in their bubble and are not exposed to different opinions and ideas.
I believe that Web 2.0 / whole user generated content thing was kinda mistake, or maybe "we" (or actually you, oldas) weren't prepared for its consequences.
Not in the first years of its existence, but in last like 10 years.
Yes, I'm aware that HN's like that, but if the price for fixing that whole mess is this small, then go ahead.
Not in the first years of its existence, but in last like 10 years.
Yes, I'm aware that HN's like that, but if the price for fixing that whole mess is this small, then go ahead.
I’d say that the ad-revenue model.
I’ve written about this here: https://medium.com/anti-content/the-shape-of-the-internet-34...
I’ve written about this here: https://medium.com/anti-content/the-shape-of-the-internet-34...
Two things discovery (search) and dns (centralized). Not sure if they are both the same problem.
We sort of don’t have an Internet Bill of Rights.
We have a patient bill of rights for example.
We have a patient bill of rights for example.
Phishing emails that take users to fake sites where they are asked to login resulting in stolen passwords. This insecurity makes users less likely to try new sites for fear of malicious JavaScript etc.
Email assumes that I want to see everything addressed to me. I don't so I have a spam/junk folder. But a lot of resources are wasted sending content that people don't care about.
Email should become a pull system, where you get a notification that a message is waiting for you on server X. Server X is trusted? Then I'll take a look, oh, it's fakegoogle@serverx, I'll block them specifically. I never see that content again, period. No one else can put a message into fakegoogle@serverx's outbox so I can be confident that I'm not accidentally blocking valid content (no address spoofing). If I trust someone, like [email protected], then I can automatically retrieve the content until you prove untrustworthy (either by your deliberate actions, or by having weak security that grants others access to your system).
This also handles not just spam, but promotional content. If someone wants to send out promotional content to 1 million customers, I can optionally retrieve it or just "unsubscribe" by never retrieving it. It's on them to continue hosting content that's never retrieved so at some point they'll detect that Jtsummers has stopped getting their content and just stop sending it to me.
Email should become a pull system, where you get a notification that a message is waiting for you on server X. Server X is trusted? Then I'll take a look, oh, it's fakegoogle@serverx, I'll block them specifically. I never see that content again, period. No one else can put a message into fakegoogle@serverx's outbox so I can be confident that I'm not accidentally blocking valid content (no address spoofing). If I trust someone, like [email protected], then I can automatically retrieve the content until you prove untrustworthy (either by your deliberate actions, or by having weak security that grants others access to your system).
This also handles not just spam, but promotional content. If someone wants to send out promotional content to 1 million customers, I can optionally retrieve it or just "unsubscribe" by never retrieving it. It's on them to continue hosting content that's never retrieved so at some point they'll detect that Jtsummers has stopped getting their content and just stop sending it to me.
Pull might also be a nice way to address the 'right now!' problem. Some, likely most information can wait till you're ready and doesn't need be available till then.
Yes, would love a 'pull' based system. You can usually tell just from the subject/preview whether you actually need an email.
There trying to make end to end encryption illegal
I think the consolidation of the internet has made it a lot less fun. I really miss personal websites being a popular thing.
the "modern" internet - web x.0 - has a focus on ads, (personal) data-collection and commercialization.
this is a social not a technical problem -> imho. its broken beyond repair.
this is a social not a technical problem -> imho. its broken beyond repair.
Ads and laws
Social media
https://builders.mozilla.community/ https://mozillabuilders.slack.com/