I don't think advocating people make their entire website out of javascript, a very "big web" requirement, is smart. It just seems like they're doing so to promote their particular npm based product and cause centralization in it.
No. Small web is your own server, yes, but also your own HTML. Small web is when a webpage is a webpage and not an application that is impenetrably dense and non-modular and frankly too complex for most people to edit. Small web is a static site that doesn't have exploits, doesn't require package managers, and lasts forever. Anyone can, and did (back in the 90s) make simple functional websites that had text and images.
This site.js is not small web and I really hope he reconsiders pushing this on small web curious noobs who it will hurt.
The privacy policies, legal terms, GDPR clickthroughs etc. all came about because everyone was already getting so evil with the cookies and tracking etc.
It’s nice to think they all cl appeared together as symptoms of Big Web but the truth is, the very early web that was built on trust and collaboration only lasted a blink of an eye before the bs starts to creep in.
I laughed out loud when I got to the "...and that's why we're pushing a new javascript framework" bit.
JS is the asphalt for the race-to-the-bottom of frontend fashion and maximizing engagements, i.e. "The Big Web". It is not required for small, functional websites, and indeed tends to make them less small and less functional/maintainable.
I don't see how Site.js is any different than what he describes as "big web". So you create your own website and host it on a server. That's always been possible. What am I missing?
I think "Big Web" is a profile hosted on "cloud infrastructure" owned by a company with public stock. Whereas "small web" is a website you build/create yourself and host on your own server?
This tracks with me until I consider the server/client rapport. As far as I can read into the author's intent, this dynamic does not change when someone hosts their own server.
If anything, the "dumb delivery mechanism" only became slightly more trustworthy?
But what is "your own server"? Yeah I host my personal site on my raspberry pi, but most people aren't going to do that. They're going to host on a server owned by a company with public stock.
It seems to me like one could market a device that isn't much more than a RPI that would download a container image from small-web-dot-registry that would run a webserver that encapsulates the site one authored on small-web-dot-com.
This image could have some smarts to either punch the right holes through the the router's firewall or do something even more clever with a peer-to-peer mesh network.
At the end of the day, what one would have is a website of their own that they control within their residence. It would foster a lot of this small web spirit and be very cheap to bring to market and maintain.
Look at the Great Leap Forward. Ideologues see everything which dares to disagree with them in even an aesthetic sense as ideological while the target wonders what the hell are they talking about. To be frank no matter how well educated they may be they are fundamentally complete idiots because of their insensitivity to reality.
I share the same ideals as mentioned in this article, but implementation is what matters. The person who wrote this article references mastodon as an example of the small web, but just looking at mastodon, you can see that is not what happens. Mastodon has a couple of very large hosts with thousands or hundreds of thousands of users, exactly the same as the big web that the author describes. This is because people think "well, why go to the trouble of setting all of this stuff up myself, I will just create an account on one of the existing hosts", and then you just have the big web again.
I think there is a huge difference between deciding to sign up on a huge host vs having to use the big host.
Mastodon maybe isn't the best example of this as IIUC you will get some visibility to other users on your host by default. But for example email you may choose to use GMail or Outlook because it takes no effort, but there is little downside to choosing another host.
On Mastodon you can choose among dozens of servers, you can decide to change server and still be able to use Mastodon. Now compare that with Youtube or Twitter. You can fix some aspects using browser extensions[1] or clients[2] but by and large, if you don't like something about the platform, the only available choice is leaving.
This is so painfully bullshit. Imagine the scientists and engineers at bell labs thinking that the operators of their machines are akin to drug addicts.
The common quip is from the mid 90s, usually attributed to Edward Tufte but apparently the real source is "a guy from Wired":
> And o there is this incredible remark which a guy from Wired told me, that there are only two industries which refer to their customers as users, drugs and computers. "I've just got to have release 7.2", right!
Well the drug dealers were certainly there first, etymonline.com has this to say of "user": Of narcotics, from 1935; of computers, from 1967."
I am struggling to find a source for who first called users "users" back in 1967 - Actually I see it being used in the 1965 "Multics" papers [0] and the 1964 article in The Atlantic "Computers of Tomorrow" [1], as if it was already a term-of-art.
OED gives the following usages, the first of which suggest users predate computers:
1950 Science 112 732/1 Analog machines..are enthusiastically supported by their users.
1959 E. M. McCormick Digital Computer Primer x. 139 The number of instructions which can be executed by a computer represents a compromise between the designer's and user's requirements.
Thanks for a good reference! It makes sense that "users of a service" or "user of a machine" slipped its way into computer jargon, no need to be coined or anything.
I'm tempted to pick up a copy of Digital Computer Primer on ebay, I love the old explanations of what these machines even are.
"Users" was a common term for people at PCs in my early networking days (Novell, early '90s), so it's actually much older than the Web. We made jokes about it being the same term used for junkies, but we certainly never meant it in that context. The people who were "addicted" were the ones running the network; the users were happy to leave it all at the end of the day.
That's why I haven't bought a console in a decade! It's such a huge time sink when you add it all up; virtual chains that you wear willingly.
That's part of why the metaverse is such dystopian bullshit. Like I need my entire field of view taken away too, so Lord Zuck can sell more ads? Ew, gross, no.
Same reaction. At first I thought I was reading something about small websites and wresting control from centralized cloud and social media companies, and suddenly it turns into something about some JS package? Tangling with anything JS does not strike me as the simplest way to host your own small site.
A 1996 definition and a 2003 version of that definition.
151 "user" jargon "The Jargon File (version 4.4.7, 29 Dec 2003)"
user
n.
1. Someone doing "real work" with the computer, using it as a means rather than an end. Someone who pays to use a computer. See {real user}.
2. A programmer who will believe anything you tell him. One who asks silly questions. [GLS observes: This is slightly unfair. It is true that users ask questions (of necessity). Sometimes they are thoughtful or deep. Very often they are annoying or downright stupid, apparently because the user failed to think for two seconds or look in the documentation before bothering the maintainer.] See {luser}.
3. Someone who uses a program from the outside, however skillfully, without getting into the internals of the program. One who reports bugs instead of just going ahead and fixing them.
The general theory behind this term is that there are two classes of people who work with a program: there are implementors (hackers) and {luser}s. The users are looked down on by hackers to some extent because they don't understand the full ramifications of the system in all its glory. (The few users who do are known as real winners.) The term is a relative one: a skilled hacker may be a user with respect to some program he himself does not hack. A LISP hacker might be one who maintains LISP or one who uses LISP (but with the skill of a hacker). A LISP user is one who uses LISP, whether skillfully or not. Thus there is some overlap between the two terms; the subtle distinctions must be resolved by context.
.
151 "user" foldoc "The Free On-line Dictionary of Computing (30 December 2018)"
user
1. <person> Someone doing "real work" with the computer, using it as a means rather than an end. Someone who pays to use a computer. A programmer who will believe anything you tell him. One who asks silly questions without thinking for two seconds or looking in the documentation. Someone who uses a program, however skillfully, without getting into the internals of the program. One who reports {bugs} instead of just fixing them. See also {luser}, {real user}.
Users are looked down on by {hackers} to some extent because they don't understand the full ramifications of the system in all its glory. The term is relative: a skilled hacker may be a user with respect to some program he himself does not hack. A LISP hacker might be one who maintains LISP or one who uses LISP (but with the skill of a hacker). A LISP user is one who uses LISP, whether skillfully or not. Thus there is some overlap between the two terms; the subtle distinctions must be resolved by context.
2. <jargon> Any person, organisation, process, device, program, {protocol}, or system which uses a service provided by others.
The term "{client}" (as in "{client-server}" systems) is rather more specific, usually implying two processes communicating via some protocol.
The thing about users was cringy, but otherwise I like this. I am not sure where to put it though, should I power up Raspberry Pi for this and let it run all the time or really use my desktop.
Maybe something distributed would be good idea. Then you can use your phone to update it. But then you get into whole crypto thing way too much?
I think this is useful thinking and exploration and if we suspend skepticism, it might be fun. I don't like seeing Twitter and FB and Youtube censoring people (and morons) all the time. I believe in the right of morons to publish stupid things.
I think the ideals are similar to indie web [https://indieweb.org/], so maybe the author may want to join that community. Basically the indie web feels like a mature community-based approach to what the author is describing, minus the javascript library.
My thoughts exactly. The IndieWeb community has been around for some time.
This article has been discussed before, and people familiar with IndieWeb made the same point back then. [0] The author responded [1] but in my opinion it was weak sauce.
Also, IndieWeb is a much better label than small web.
> Site.js is now the easiest way to set up and run your own Owncast instance. This will install Owncast, set it up as a systemd service, and serve it securely at your hostname.
Unfortunately, 99.99% of people this is meant for, that are currently users of big web, have no idea what the two previous sentences mean.
but hopefully leaving people not as clueless. But to not learn a thing seems to be part of the long-term promise. While short term they have to be devs.
I would like that the way round. Start with laypersons to begin with. That may not look high-tech, however. Hard to get tech-funding for such.
> The Small Web is for people (not startups, enterprises, or governments). It is also made by people and small, independent organisations (not startups, enterprises, or governments2).
Coincidentally the Who is hiring is front page today. In the comments do a CTRL+F and search for TypeScript and see how many posts are directly correlated with a ginormous framework like React, Angular, or Vue.
Like many here I want to dig the spirit of this but I am confused about the concrete steps / tools they're describing.
It always seemed to me like ultimately the problem with getting free of the big web would be the physical infrastructure side of the equation.
The modern web does feel controlled by information age rentiers but growing up in the 90s that always felt like an unavoidable given to me.
I was under the impression that the freedom of the web lay in the freedom of information and the subsequent empowerment of individuals to maneuver through and undermine larger systems of power.
While I'm not sure about some of the half-said political attitude in this post, I do strongly believe in what the author is getting at, and am also building a company for "the Small Web". 1
I've been working on an essay for quite awhile now on a similar topic: what does a decentralized web actually look like? If Big Cloud is an empire-sized industry and Blockchain is a wildly re-engineered approach to solving the centralized app problem, then the simple dark-horse is just more folks running server software (at home or just on computers they own). TCP/IP and HTTP as-is decentralize just fine - we have an education and tooling problem, not a technical problem.
> The spiders that sit at its centre waiting to suck you dry are Big Tech people farmers like Facebook, Google
Look, Big Tech doesn't have to be evil for there to be advantages to running your own software. There are _much more immediate_ benefits to every day people, like owning all the rights to all the pictures of their children, seeing fewer ads when consuming media, or having a way to host a website without paying a fortune 10 company a monthly stipend. Melodrama and angry politics aside, we at least agree that it's time to fix the internet :)
Small Web applications and sites are single tenant. That means that one server hosts one application that serves just one person: you.
I feel like people forgot that shared hosting exists (e.g. Dreamhost, or Nearly Free Speech).
One computer can serve tons of websites! It can serve multiple Hacker News spikes at once. And the site doesn't even need to be in a virtual machine or container.
This is how "LAMP" / PHP works! But these days it works with Python too. Either way it's perfect for static and JS-only sites.
A single computer is often more reliable than a cloud. The cloud has single points of failure that are less reliable than a commodity machine.
I think the problem is that shared hosting has a shell interface (this used to be a desirable feature!!). And many users no longer know shell. Hopefully we can make it easier to learn!
The biggest problem with shared hosting is security. It's common for everyone's processes to run as the same "www-data" user. It can be made secure, but it isn't common and most shared hosting is powered by off-the-shelf piles of shit like cPanel which are Swiss cheese when it comes to security.
No I don't believe any shared hosting does this (at least not ones anybody pays money for). Dreamhost and Nearly Free Speech definitely don't.
This might have been true 20 years ago, but I doubt it was true 10 years ago, and definitely not today. Maybe some tiny universities had janky setups like this.
Fun fact: shared hosting providers maintained OS virtualization for Linux years before Docker existed.
I'm not quite sure but it also appears to have pre-dated cgroups. If you pack many users together on a server, you will want something like cgroups to prevent neighbors from DOS-ing each other.
This was a fork and wasn't in mainline Linux.
So they actually had MORE isolation than running as separate Unix users, not LESS.
(edit: I'm not sure exactly which shared hosting providers used such Linux kernel forks, but I'd be interested in anyone with direct knowledge. From using the shell on various shared hosting providers, I know they have pretty custom configurations with more isolation than stock distros.)
Nope, most shared hosting is still like the OP comment’s description.
Then of course you have companies like Cloudflare and Netlify where their shared static hosting is not the traditional lamp/cpanel stuff, and certainly have internally developed platforms/serving technologies where security and sandboxing is very likely top of mind.
Companies of Cloudflare and Netlify ilk are the exception, but most cheapo shared hosting (and shared hosting has to be cheap to compete), is still very much multi-tenant with security concerns across the board.
I've used 3 shared hosting providers, all of which provide SSH access. And when I ssh in, and type "ps", I see my own processes, not the processes of anybody else. They are running as a different user.
You can have unique SSH users and still have all the PHP code run under a shared user on the web server. The machine you SSH in is most likely not the same as the one running your PHP code even.
I can't imagine that the process level is an issue here – we're not talking military, do we?
Other than that, prefer zero-knowledge designs and you can often reduce to the danger of vandalism. Have a friendly hoster and/or backups for that case.
While not military, most websites handle some kind of personal data (even IP addresses count as such for GDPR purposes) and a large chunk of the world has regulations that impose penalties or at the very least requires disclosure (bad PR) for data breaches.
Considering how easy and relatively cheap it is to get your own dedicated bare-metal server, I don't see any reason to bother with shared hosting anymore.
Yeah so I use Python and FastCGI with Dreamhost, which runs as a separate process and separate user.
I think mod_php (Apache shared library) is the the thing where every PHP site runs as www-data, but I thought that FastCGI was preferred these days over mod_apache. FastCGI is a separate process, which can run as a separate user, communicating over sockets.
But the original post is about STATIC websites, so this won't come into play. The main point is that you shouldn't spin up a VPS to serve static sites! You'll be burdened with maintaining the OS and web server.
Shared hosting is basically as easy as Github pages, with more choice of provider (in keeping with the philosophy of the small web).
These days shared hosting with multiple tenants is more commonly known as "serverless". A bit of a misnomer since servers are involved, but it's the common term.
another thing I like is stuff like "lambda" where you can run a function instead of a whole server application for every api request! Wait, I think I've seen this before.
functionname.cgi?param1=val¶m2=val
(replace "cgi" with your favourite php/pl/py/cfm/whatever)
Differences between shared hosting and "serverless":
- Monthly or yearly billing vs hourly (as the sibling mentioned). I think my plan is $9/month or something, plus $16/year for domain name.
- Single server vs. multiple servers. AWS Lambda is great if you actually need to scale up, but you don't need that for serving a website. A single machine can serve hundreds of static websites.
- You can get commodity shared hosting from dozens of providers, but you can get serverless from only a few "big cloud" providers, which goes against the ethos of the small web. That's because writing and maintaining the custom software to scale up is a big undertaking.
The original post is talking about static websites. You don't need anything running on the server for this! So all you need is shared hosting, and you won't even use the PHP part.
But agree on people not knowing shell, even cloud shell on azure/aws - people like to touch things and really enjoy the UI for setting up networking rules, load balancers and more.
Still so much fun, cgi - php and apache in general, the mid 00's were an exciting time were you could do so many things. I enjoyed hosting a website on port 80 from my residential cable/roadrunner connection and editing live pages and seeing it reflect instantly (sure you could just do localhost, but back then even port 25 and such were open, easily letting you run your own mail server and more.)
But the fastest way to get a page on the web nowadays would be github pages, you can do it in less than 10 mins, but I guess if you have a server already imaged, you can do "sudo apt-get install lamp-server" and get started that way.
Containerization solves 99% of those problems and more, such as not having to worry about conflicting dependencies, share of common resources, etc.
I would say shared hosting is now more popular than ever, with a twist: similar to the shift that has happened from multi-user mainframes to personal computers to ubiquitous computing (ie individuals having multiple computers), self-hosters are also now hosting not one but many websites/apps per server.
I am not sure how the status quo is different from your suggestion, or rather, how your suggestion is an improvement over that?
> self-hosters are also now hosting not one but many websites/apps per server
When was this not the case? Self-hosters are generally more resource-constrained, and thus reusing hardware makes more sense. It was pretty common already 20 years ago.
I shouldn't have brought up LAMP / PHP, because that confuses the issue.
The point is that you should not use a VPS to serve static content! You will not do as good a job as maintaining the OS and web server as a professional. Just use shared hosting!
You can also use Github Pages or whatnot, but that goes against the philosophy of centralized cloud providers.
Commodity shared hosting is very portable between providers, because it's just a Linux box, without "cloud" stuff. That's all you need for static pages. Use rsync or git.
Probably the same sentiment as many of the comments already present, but I haven't read all of them.
Don't people understand that the original Internet was/is built for exactly this(?); end to end connectivity between any two hosts. This is still mostly possible, but you do now have to get a public/routable ipv4 address. IPv6 gets back to the E2E ideal, without any overlay.
P2P systems from 20 years ago also solved this problem. For example hidden services on Tor, i2p, generic DHTs, etc.
Why people think we need to redo everything so that it works with JS and Blockchain boggles the mind. It's not corporate vs. individual as the seem to argue, it's just pro Blockchain, or pro Ethereum or something else.
This article seems to just be framing this same non-argument in a slightly different way. Get a static, public IP and run your microhttpd blog for pennies. Don't act like hosting it on an overcomplicated overengineed Blockchain network is going to decentalize the web. It's already decentralized, but nobody wants to go to your hidden service...
At uncharitable times I feel like the "small web" is about making a new hype machine around a certain ideology.
> P2P systems from 20 years ago also solved this problem. For example hidden services on Tor, i2p, generic DHTs, etc.
In addition it's easier than ever to get static IPv4 services or to setup a tiny VPS whose sole purpose is to offer a static IPv4 that routes traffic from/to computers on your LAN. Other than Tor hidden services, there's also tons of other internet software out there old and new for network connected devices to share messages like Email, Usenet, Bitmessage, UUCP, NNCP, Secure Scuttlebutt, IPFS, DAT, nostr, etc, etc.
This is not the solution to the problems we’re facing. While it’s nice to push people to run their own hardware and software, it does nothing to solve the problem of large corporations having large amounts of control and power with little regulation oversight.
We like to do this thing in tech where we try to solve all of our problems with engineering and software. What we’re encountering now is not something that can be solved with software: it’s a political and economic problem.
We have large corporations who have the power of small countries with little (if any) oversight on the stuff they build. Apple is attempting to scan users’ photos. Amazon is using software to exploit their workers and treat them like shit, and to stalk their users around the web into buying more shit. Facebook is giving teens eating disorders and trying to create a version of Instagram for children.
The problem we’re having now is fundamentally a political problem. We have to work together as a society, collectively instead of individually, to decide what the limits of these corporations should be. That happens through boring political processes and debates that nobody likes, which isn’t appealing to us programmers. But it’s the reality we face.
Individualistic ‘solutions’ like this only cause more harm because they don’t address the actual problems at hand. It’s like when Shell introduced the idea of the ‘carbon footprint’ to shift the conversation from us as a society holding corporations accountable to individuals being held accountable. The focus *must* be on political solutions to these problems, because there is no technical way this can be solved.
But there are technical ways this can be solved. Don't use their services is one.
I've got a lot of problems but you know what problems I don't have? Being censored and tracked by facebook and google. It really is very much an individual problem with individual solutions, though some may not like that fact because individual solutions don't accomplish their goals, it is still a fact.
Please explain to me the path from individuals hosting their own Small Web apps to the mitigation of Amazon's power and prevalence in our lives. It's nice that you are off fb, I am too. But some number of individuals deciding not to use the platform does nothing when it is the way most people access the internet. People don't like this example because of its stereotypical user base, but remember what happened to parlor? That's what FAANG does, and will do, to individual small competitors unless their power is challenged in the political and economic arena.
If I don't use amazon, they have no power over me. They're background noise, like sally beauty salon. And consequently, the less people that use them the less power they have. It's easy to forget that their power isn't immutable, we give it to them.
That's idealistic thinking. Amazon does have power over people that don't use it. It influences national, state and local governments. Where it decides to open a distribution center has a huge impact on communities. It controls AWS, which is essentially the interstate highway system of the internet. You might not perceive the effects, but that doesn't mean it has no power over you. With Amazon the logistics and e-commerce juggernaut it is, there's really no meaningful alternative, and saying "just don't use it" is not very helpful.
Please read through my comment again. The “don’t use their services” line is another individualistic framing that exists solely to deflect blame from these large corporations.
Shell launched a PR campaign in the 90’s to popularize the term ‘carbon footprint’. This shifted the climate change conversation from large, multinational corporations needing to solve it to individuals, who have little-to-no power, needing to solve it. 71% of global CO2 emissions are emitted by just 100 companies. Even if every individual on Earth lowered their output by as much as possible, we’d all still be fucked because of the emissions by these corporations.
We must talk about corporate power in terms of policy, government, and collective responsibility. Individuals have no power in comparison to these large corporations. This is something we’ve observed in nearly every market these corporations compete in: if I had a dime for every company that tried to compete with Google Search and died over the years, I’d be really fucking rich. While their might be technical alternatives to these services, that doesn’t solve the problem of these corporations having unrestricted power to do whatever they want. That’s what we need to solve.
Appreciate your take on this. Those of us who wince at spending time in the Social Mess Halls may be inclined to build our personal curios in dusty corners of the net, content with a dozen views a month, but that does not serve as a rebellion to the problems in tech. There must be some better way we can think as groups of people, on necessarily large platforms. Something better than Dorsey and Zuck have managed to build.
I am very concerned about the metaverse trend, it's like people watched the Matrix and thought "yeah, let me plug into the machine and not deal with real life" and didn't catch the moral of Ready Player One, "Real Life is the only thing thats Real", but were sold on the nostalgia orgasm of having Jurrasic Park and Back to the Future in the same game.
I would like to build some anti-meta-verse, that helps you build the actual world by keeping your attention on your own intentions, not allowing your mind to wander at the whim of ad-campaign designers (we can all agree the IOI represented facebook, right? the bad guys that want to fill 80% of your peripheral vision with ads? but if it is more addictive then real life, of course people will choose to spend their time there.)
Long term you're probably right, but I think tunneling services[0] like Cloudflare tunnel or my own boringproxy are going to be what we use for the next 10 years or so.
We're going to have to prove that self-hosting is useful before people start demanding ipv6, public static IPs, and decent upload speeds from their ISPs.
I have a little site that I moved behind cloudflare. The dead simple https is nice but I have a page with a gallery of 140 images and it loads so much slower than before the move. Granted I am not paying anything for the service and I don't plan to change it back so this is just an observation.
That's a hot take. Do you never do anything on the web which you don't want your ISP to have full access to?
That said, I agree HTTPS is too complicated to set up, even with ACME/LetsEncrypt. I wish you could just set up your own certificate with the public key in your DNS records. But that requires DNSSEC/DoH or similar.
How much traffic does the gallery get? Unless you have enough people in a given area hitting the site enough for the Cloudflare PoP in that area to have a warm cache, the first request for each image is going to be slower than going directly to the origin server.
It is a site for a game that died ten years ago so it gets more bots than people at this point. Last month my cloudflare email said they saved me 22.7% of 91.69MB.
For that use case, yep totally agree. But there are a lot of use cases that an extra hop through a proxy that's physically near the origin server still provides very good performance.
SQLite author is an avid Tcl user and he recently introduced a small, secure and modern CGI based web application called wapp [1],[2]. I think this small web concept proposal is excellent fit for wapp.
Coincidently, a few days back there was an HN post on a novel backend-less web application using SQLite for off-line friendly apps [3]. Combination of small web and local-first software will probably a new killer feature of the Internet in the near future [4].
while operating the server is the full thing, shared hosting is sensibly doable for most if not all people taking part in the internet.
It's cumbersome application stacks that make hosting a part time job and scare the hell out of laypeople. Disclaimer: I evangelize single-file static monoliths like https://demo.mro.name/shaarligo
125 comments
[ 3.0 ms ] story [ 192 ms ] threadNo. Small web is your own server, yes, but also your own HTML. Small web is when a webpage is a webpage and not an application that is impenetrably dense and non-modular and frankly too complex for most people to edit. Small web is a static site that doesn't have exploits, doesn't require package managers, and lasts forever. Anyone can, and did (back in the 90s) make simple functional websites that had text and images.
This site.js is not small web and I really hope he reconsiders pushing this on small web curious noobs who it will hurt.
It’s nice to think they all cl appeared together as symptoms of Big Web but the truth is, the very early web that was built on trust and collaboration only lasted a blink of an eye before the bs starts to creep in.
JS is the asphalt for the race-to-the-bottom of frontend fashion and maximizing engagements, i.e. "The Big Web". It is not required for small, functional websites, and indeed tends to make them less small and less functional/maintainable.
If anything, the "dumb delivery mechanism" only became slightly more trustworthy?
This image could have some smarts to either punch the right holes through the the router's firewall or do something even more clever with a peer-to-peer mesh network.
At the end of the day, what one would have is a website of their own that they control within their residence. It would foster a lot of this small web spirit and be very cheap to bring to market and maintain.
Mastodon maybe isn't the best example of this as IIUC you will get some visibility to other users on your host by default. But for example email you may choose to use GMail or Outlook because it takes no effort, but there is little downside to choosing another host.
[1] e.g. https://www.undistracted.app/
[2] e.g. https://freetubeapp.io/
This doesn't seem like it's true. I feel like it came from the unix culture. Unless they got it from drug dealers...
Oddly relevant to social media nowadays though.
> And o there is this incredible remark which a guy from Wired told me, that there are only two industries which refer to their customers as users, drugs and computers. "I've just got to have release 7.2", right!
[0] https://cryptome.org/jya/tufte.htm
I am struggling to find a source for who first called users "users" back in 1967 - Actually I see it being used in the 1965 "Multics" papers [0] and the 1964 article in The Atlantic "Computers of Tomorrow" [1], as if it was already a term-of-art.
[0] https://multicians.org/history.html
[1] https://www.sas.upenn.edu/~traister/greenbf.html
1950 Science 112 732/1 Analog machines..are enthusiastically supported by their users.
1959 E. M. McCormick Digital Computer Primer x. 139 The number of instructions which can be executed by a computer represents a compromise between the designer's and user's requirements.
I'm tempted to pick up a copy of Digital Computer Primer on ebay, I love the old explanations of what these machines even are.
That's part of why the metaverse is such dystopian bullshit. Like I need my entire field of view taken away too, so Lord Zuck can sell more ads? Ew, gross, no.
2. A programmer who will believe anything you tell him. One who asks silly questions. [GLS observes: This is slightly unfair. It is true that users ask questions (of necessity). Sometimes they are thoughtful or deep. Very often they are annoying or downright stupid, apparently because the user failed to think for two seconds or look in the documentation before bothering the maintainer.] See {luser}.
3. Someone who uses a program from the outside, however skillfully, without getting into the internals of the program. One who reports bugs instead of just going ahead and fixing them.
The general theory behind this term is that there are two classes of people who work with a program: there are implementors (hackers) and {luser}s. The users are looked down on by hackers to some extent because they don't understand the full ramifications of the system in all its glory. (The few users who do are known as real winners.) The term is a relative one: a skilled hacker may be a user with respect to some program he himself does not hack. A LISP hacker might be one who maintains LISP or one who uses LISP (but with the skill of a hacker). A LISP user is one who uses LISP, whether skillfully or not. Thus there is some overlap between the two terms; the subtle distinctions must be resolved by context.
1. <person> Someone doing "real work" with the computer, using it as a means rather than an end. Someone who pays to use a computer. A programmer who will believe anything you tell him. One who asks silly questions without thinking for two seconds or looking in the documentation. Someone who uses a program, however skillfully, without getting into the internals of the program. One who reports {bugs} instead of just fixing them. See also {luser}, {real user}.Users are looked down on by {hackers} to some extent because they don't understand the full ramifications of the system in all its glory. The term is relative: a skilled hacker may be a user with respect to some program he himself does not hack. A LISP hacker might be one who maintains LISP or one who uses LISP (but with the skill of a hacker). A LISP user is one who uses LISP, whether skillfully or not. Thus there is some overlap between the two terms; the subtle distinctions must be resolved by context.
2. <jargon> Any person, organisation, process, device, program, {protocol}, or system which uses a service provided by others.
The term "{client}" (as in "{client-server}" systems) is rather more specific, usually implying two processes communicating via some protocol.
This is a JavaScript library that runs sites on a rented server. Might as well be in the cloud.
Maybe something distributed would be good idea. Then you can use your phone to update it. But then you get into whole crypto thing way too much?
I think this is useful thinking and exploration and if we suspend skepticism, it might be fun. I don't like seeing Twitter and FB and Youtube censoring people (and morons) all the time. I believe in the right of morons to publish stupid things.
This article has been discussed before, and people familiar with IndieWeb made the same point back then. [0] The author responded [1] but in my opinion it was weak sauce.
Also, IndieWeb is a much better label than small web.
[0] https://news.ycombinator.com/item?id=24268472
[1] https://news.ycombinator.com/item?id=24269830
Unfortunately, 99.99% of people this is meant for, that are currently users of big web, have no idea what the two previous sentences mean.
I would like that the way round. Start with laypersons to begin with. That may not look high-tech, however. Hard to get tech-funding for such.
Coincidentally the Who is hiring is front page today. In the comments do a CTRL+F and search for TypeScript and see how many posts are directly correlated with a ginormous framework like React, Angular, or Vue.
It always seemed to me like ultimately the problem with getting free of the big web would be the physical infrastructure side of the equation.
The modern web does feel controlled by information age rentiers but growing up in the 90s that always felt like an unavoidable given to me.
I was under the impression that the freedom of the web lay in the freedom of information and the subsequent empowerment of individuals to maneuver through and undermine larger systems of power.
Connects to your wifi like your smart tv does, and then does some black magic to get around the NAT issue to register a subdomain that you can use.
The device provides a facebook or twitter like UI to set up your site.
Even then Im not sure how you'll get people off social media onto this as it is still less convenient unfortunately.
https://freedombox.org/buy/
It's a bit like rules for being a punk rocker, no?
I've been working on an essay for quite awhile now on a similar topic: what does a decentralized web actually look like? If Big Cloud is an empire-sized industry and Blockchain is a wildly re-engineered approach to solving the centralized app problem, then the simple dark-horse is just more folks running server software (at home or just on computers they own). TCP/IP and HTTP as-is decentralize just fine - we have an education and tooling problem, not a technical problem.
> The spiders that sit at its centre waiting to suck you dry are Big Tech people farmers like Facebook, Google
Look, Big Tech doesn't have to be evil for there to be advantages to running your own software. There are _much more immediate_ benefits to every day people, like owning all the rights to all the pictures of their children, seeing fewer ads when consuming media, or having a way to host a website without paying a fortune 10 company a monthly stipend. Melodrama and angry politics aside, we at least agree that it's time to fix the internet :)
1 https://kubesail.com
What if I want no server, and websites I want to read is not about people.
Small Web applications and sites are single tenant. That means that one server hosts one application that serves just one person: you.
I feel like people forgot that shared hosting exists (e.g. Dreamhost, or Nearly Free Speech).
One computer can serve tons of websites! It can serve multiple Hacker News spikes at once. And the site doesn't even need to be in a virtual machine or container.
This is how "LAMP" / PHP works! But these days it works with Python too. Either way it's perfect for static and JS-only sites.
A single computer is often more reliable than a cloud. The cloud has single points of failure that are less reliable than a commodity machine.
----
I mentioned that here:
What's the Fastest Way To Get a Page on the Web?
https://news.ycombinator.com/item?id=29254006
http://www.oilshell.org/share/question.html
http://www.oilshell.org/blog/2021/12/backlog-assess.html#wha...
I think the problem is that shared hosting has a shell interface (this used to be a desirable feature!!). And many users no longer know shell. Hopefully we can make it easier to learn!
This might have been true 20 years ago, but I doubt it was true 10 years ago, and definitely not today. Maybe some tiny universities had janky setups like this.
Fun fact: shared hosting providers maintained OS virtualization for Linux years before Docker existed.
https://en.wikipedia.org/wiki/OpenVZ -- initial release 2005
I'm not quite sure but it also appears to have pre-dated cgroups. If you pack many users together on a server, you will want something like cgroups to prevent neighbors from DOS-ing each other.
This was a fork and wasn't in mainline Linux.
So they actually had MORE isolation than running as separate Unix users, not LESS.
(edit: I'm not sure exactly which shared hosting providers used such Linux kernel forks, but I'd be interested in anyone with direct knowledge. From using the shell on various shared hosting providers, I know they have pretty custom configurations with more isolation than stock distros.)
This page is a little fuzzy: https://en.wikipedia.org/wiki/Virtuozzo_(company)
Then of course you have companies like Cloudflare and Netlify where their shared static hosting is not the traditional lamp/cpanel stuff, and certainly have internally developed platforms/serving technologies where security and sandboxing is very likely top of mind.
Companies of Cloudflare and Netlify ilk are the exception, but most cheapo shared hosting (and shared hosting has to be cheap to compete), is still very much multi-tenant with security concerns across the board.
https://news.ycombinator.com/item?id=29804968
I've used 3 shared hosting providers, all of which provide SSH access. And when I ssh in, and type "ps", I see my own processes, not the processes of anybody else. They are running as a different user.
Other than that, prefer zero-knowledge designs and you can often reduce to the danger of vandalism. Have a friendly hoster and/or backups for that case.
Considering how easy and relatively cheap it is to get your own dedicated bare-metal server, I don't see any reason to bother with shared hosting anymore.
My bet would be that most larger shared hosts have implemented some type of solution for this, but perhaps some smaller ones haven't.
I think mod_php (Apache shared library) is the the thing where every PHP site runs as www-data, but I thought that FastCGI was preferred these days over mod_apache. FastCGI is a separate process, which can run as a separate user, communicating over sockets.
But the original post is about STATIC websites, so this won't come into play. The main point is that you shouldn't spin up a VPS to serve static sites! You'll be burdened with maintaining the OS and web server.
Shared hosting is basically as easy as Github pages, with more choice of provider (in keeping with the philosophy of the small web).
https://en.m.wikipedia.org/wiki/Serverless_computing
Shared hosting is usually much more straightforward in many ways, including flat-rate pricing.
- Monthly or yearly billing vs hourly (as the sibling mentioned). I think my plan is $9/month or something, plus $16/year for domain name.
- Single server vs. multiple servers. AWS Lambda is great if you actually need to scale up, but you don't need that for serving a website. A single machine can serve hundreds of static websites.
- You can get commodity shared hosting from dozens of providers, but you can get serverless from only a few "big cloud" providers, which goes against the ethos of the small web. That's because writing and maintaining the custom software to scale up is a big undertaking.
The original post is talking about static websites. You don't need anything running on the server for this! So all you need is shared hosting, and you won't even use the PHP part.
https://news.ycombinator.com/item?id=29804920
But agree on people not knowing shell, even cloud shell on azure/aws - people like to touch things and really enjoy the UI for setting up networking rules, load balancers and more.
Still so much fun, cgi - php and apache in general, the mid 00's were an exciting time were you could do so many things. I enjoyed hosting a website on port 80 from my residential cable/roadrunner connection and editing live pages and seeing it reflect instantly (sure you could just do localhost, but back then even port 25 and such were open, easily letting you run your own mail server and more.)
But the fastest way to get a page on the web nowadays would be github pages, you can do it in less than 10 mins, but I guess if you have a server already imaged, you can do "sudo apt-get install lamp-server" and get started that way.
> This blog post was published on April 1. That’s a huge hint ;-)
Not sure what to think about this...
I would say shared hosting is now more popular than ever, with a twist: similar to the shift that has happened from multi-user mainframes to personal computers to ubiquitous computing (ie individuals having multiple computers), self-hosters are also now hosting not one but many websites/apps per server.
I am not sure how the status quo is different from your suggestion, or rather, how your suggestion is an improvement over that?
When was this not the case? Self-hosters are generally more resource-constrained, and thus reusing hardware makes more sense. It was pretty common already 20 years ago.
And mentions a VPS:
Sync and deploy to your own VPS.
I shouldn't have brought up LAMP / PHP, because that confuses the issue.
The point is that you should not use a VPS to serve static content! You will not do as good a job as maintaining the OS and web server as a professional. Just use shared hosting!
You can also use Github Pages or whatnot, but that goes against the philosophy of centralized cloud providers.
Commodity shared hosting is very portable between providers, because it's just a Linux box, without "cloud" stuff. That's all you need for static pages. Use rsync or git.
This page looks really ugly but it will give you a sense of how many independent providers there are: https://www.webhostingtalk.com/forumdisplay.php?f=4
Don't people understand that the original Internet was/is built for exactly this(?); end to end connectivity between any two hosts. This is still mostly possible, but you do now have to get a public/routable ipv4 address. IPv6 gets back to the E2E ideal, without any overlay.
P2P systems from 20 years ago also solved this problem. For example hidden services on Tor, i2p, generic DHTs, etc.
Why people think we need to redo everything so that it works with JS and Blockchain boggles the mind. It's not corporate vs. individual as the seem to argue, it's just pro Blockchain, or pro Ethereum or something else.
This article seems to just be framing this same non-argument in a slightly different way. Get a static, public IP and run your microhttpd blog for pennies. Don't act like hosting it on an overcomplicated overengineed Blockchain network is going to decentalize the web. It's already decentralized, but nobody wants to go to your hidden service...
Is this redoing it? I don' get that impression. It seems ther're just making it easy, as easy as installing an app and signing up for Facebook.
> P2P systems from 20 years ago also solved this problem. For example hidden services on Tor, i2p, generic DHTs, etc.
In addition it's easier than ever to get static IPv4 services or to setup a tiny VPS whose sole purpose is to offer a static IPv4 that routes traffic from/to computers on your LAN. Other than Tor hidden services, there's also tons of other internet software out there old and new for network connected devices to share messages like Email, Usenet, Bitmessage, UUCP, NNCP, Secure Scuttlebutt, IPFS, DAT, nostr, etc, etc.
This is not the solution to the problems we’re facing. While it’s nice to push people to run their own hardware and software, it does nothing to solve the problem of large corporations having large amounts of control and power with little regulation oversight.
We like to do this thing in tech where we try to solve all of our problems with engineering and software. What we’re encountering now is not something that can be solved with software: it’s a political and economic problem.
We have large corporations who have the power of small countries with little (if any) oversight on the stuff they build. Apple is attempting to scan users’ photos. Amazon is using software to exploit their workers and treat them like shit, and to stalk their users around the web into buying more shit. Facebook is giving teens eating disorders and trying to create a version of Instagram for children.
The problem we’re having now is fundamentally a political problem. We have to work together as a society, collectively instead of individually, to decide what the limits of these corporations should be. That happens through boring political processes and debates that nobody likes, which isn’t appealing to us programmers. But it’s the reality we face.
Individualistic ‘solutions’ like this only cause more harm because they don’t address the actual problems at hand. It’s like when Shell introduced the idea of the ‘carbon footprint’ to shift the conversation from us as a society holding corporations accountable to individuals being held accountable. The focus *must* be on political solutions to these problems, because there is no technical way this can be solved.
I've got a lot of problems but you know what problems I don't have? Being censored and tracked by facebook and google. It really is very much an individual problem with individual solutions, though some may not like that fact because individual solutions don't accomplish their goals, it is still a fact.
Shell launched a PR campaign in the 90’s to popularize the term ‘carbon footprint’. This shifted the climate change conversation from large, multinational corporations needing to solve it to individuals, who have little-to-no power, needing to solve it. 71% of global CO2 emissions are emitted by just 100 companies. Even if every individual on Earth lowered their output by as much as possible, we’d all still be fucked because of the emissions by these corporations.
We must talk about corporate power in terms of policy, government, and collective responsibility. Individuals have no power in comparison to these large corporations. This is something we’ve observed in nearly every market these corporations compete in: if I had a dime for every company that tried to compete with Google Search and died over the years, I’d be really fucking rich. While their might be technical alternatives to these services, that doesn’t solve the problem of these corporations having unrestricted power to do whatever they want. That’s what we need to solve.
I am very concerned about the metaverse trend, it's like people watched the Matrix and thought "yeah, let me plug into the machine and not deal with real life" and didn't catch the moral of Ready Player One, "Real Life is the only thing thats Real", but were sold on the nostalgia orgasm of having Jurrasic Park and Back to the Future in the same game.
I would like to build some anti-meta-verse, that helps you build the actual world by keeping your attention on your own intentions, not allowing your mind to wander at the whim of ad-campaign designers (we can all agree the IOI represented facebook, right? the bad guys that want to fill 80% of your peripheral vision with ads? but if it is more addictive then real life, of course people will choose to spend their time there.)
Dynamic DNS is a hardcoded rentseeking joke in most routers, they never wanted an open standard for that.
I'm making my own DynDNS but then I now need cloud instances, which is a dependency we should not need.
Act accordingly in your local jurisdiction.
We're going to have to prove that self-hosting is useful before people start demanding ipv6, public static IPs, and decent upload speeds from their ISPs.
[0]: https://github.com/anderspitman/awesome-tunneling
That said, I agree HTTPS is too complicated to set up, even with ACME/LetsEncrypt. I wish you could just set up your own certificate with the public key in your DNS records. But that requires DNSSEC/DoH or similar.
I use this for security: https://datatracker.ietf.org/doc/html/rfc2289
Trying to avoid MitM attacks from your ISPs is really an exercise i futility, and a really expensive one at that, both energy and timevise.
HTTP, DNS and SMTP are the only protocols you need.
Small web uses small protocols, big web uses big protocols.
So you are correct that the cache is never warm.
Coincidently, a few days back there was an HN post on a novel backend-less web application using SQLite for off-line friendly apps [3]. Combination of small web and local-first software will probably a new killer feature of the Internet in the near future [4].
[1] Wapp - A Web-Application Framework for TCL:
https://wapp.tcl.tk/home
[2] EuroTcl2019: Wapp - A framework for web applications in Tcl (Richard Hipp):
https://www.youtube.com/watch?v=nmgOlizq-Ms
[3] Using the SQLite-over-HTTP “hack” to make back end-less, offline-friendly apps:
https://news.ycombinator.com/item?id=29758613
[4] Local-First Software: You Own Your Data, in spite of the Cloud:
https://martin.kleppmann.com/papers/local-first.pdf
It's cumbersome application stacks that make hosting a part time job and scare the hell out of laypeople. Disclaimer: I evangelize single-file static monoliths like https://demo.mro.name/shaarligo