Short URLs: Why and How

I don't like this at all - there is value in a hierarchical naming structure - even if it doesn't represent actual directories.

A simple example is:

https://www.kozubik.com/

... where you can find "items":

https://www.kozubik.com/items/

... and one thing inside of "items" is an article on NDS emitters:

https://www.kozubik.com/items/nds/

... which contains supporting multimedia objects:

https://www.kozubik.com/items/nds/images/

Not only do you know where you are in the tree, but it is also discoverable: by using index.html default pages and your web servers directory indexing code, you can have a very flexible resource that is easy to navigate.

I believe this is a rather naive posts on the subject of URLs and honestly wish it never made it to HN front-page, but seeing as we are here... lets go through the points made and how they conflict with each other...

  "You can remember them"

Well, no. You might remember a few. But you are only going to remember so much. This has a very limited benefit.

  "You can tell someone"

Yeah, but no. I barely tell people a domain. I'll just send them a link on the spot. Again, very limited benefit.

  "They look nicer."

Ok, but not a strong use-case. In general usage the display name of a link does not need to reflect the URL of the link.

  "They remove the middle-man" ... "encourage people to copy and paste" ...

I'm just gonna say this is plain false as a generalization. I've worked on URL sharing in publishing for almost 2 decades (yikes).

It's built on a little bit of truth. On mobile devices all have built in share functions to text and spread based on various apps installed.

Again, very limited impact and effect.

  "They’re enough."

This is my favorite! He talks about unique combinations but this goes against his original point of making them memorable. He should of at least talked about potential dictionary level word combinations... sigh

I own a 3-letter domain with a 2-letter TLD and still spells my family name. Many of my friends finds it cool when I share links via that domain. The links are indeed easy to remember but it comes at a higher yearly cost of renewal and I had forgotten to renew twice. Luckily, I was able to re-register them (nobody wants it, ;-)).

Example https://oin.am/rr/

How many people type the whole URL and how many access it over Google, QR code, or from other site? This is good if content is unique. I find URLs with dates useful when I want to quickly assess how fresh is the content.

PG's website pretty much does this

e.g. from 2003: http://www.paulgraham.com/hundred.html

most recent post in 2022: http://www.paulgraham.com/heresy.html

I'm not sure I would want to have all my writing in a single directory, but I guess you can't argue with 20 years of longevity ... on both sivers' site and PG's.

Depending on how much your write, URL collisions can naturally become a real problem. If you use a SSG then all your post HTML documents will be in the same directory, forever. This isn't a problem if you don't spit out posts, but if you post a lot your directory can become rather unwieldy. I'd say, at least throw a year in there: /2022/su/.

URLs generally do not matter to people.

Quick, what’s the URL of this discussion? Will you remember it? Does that have any impact on your usage of HN?

The vast, vast majority of website sessions start with a click. When people bother to type things in, they type them into search bars (reminder that in most browsers, the “address bar” is also the search bar).

So be clear what you want from your URL structure. If you want search traffic, you’d do better to organize it for Google (hierarchical topic-based) than making it short.

If you want to be able to say it or print it for easy typing, a domain name is going to be best. Catch it and redirect it to the target page.

Even “domain.com/word” is going to be hard for most people to remember. If they remember it at all, they will probably type in “domain word” and let Google figure it out for them.

I’m getting old because this reminds me of a time when I built a Url shortener years back. We needed about 30m unique short urls and at the time 3rd parties were expensive and a bit annoying for our use case. The project was successful but a few bumps in the road were encountered ie when we came up with the simple solution of the urls being their autoincrement IDs base…i think we used 40?…encoded we didn’t consider that sometimes the results might be offensive words. These urls were being used for referral links, and sending someone www.shorturl.com/cunt didn’t go down so well. We essentially had to sensor certain integers out to prevent this.

I do think this is a good idea in some situations, but the author doesn't get into the advantages of the style they dislike. Let's look at that path:

    /blog/2022/05/08/short-urls-why-and-how.html

I can tell it's a blog post, I can tell when it was written, and I can see enough of the title that I can remember if I already read it. The ".html" doesn't add anything though.

Personally, I find putting the title of the post (but not the date) in the path is a good compromise:

    /short-urls-why-and-how

I really appreciate that Blogspot and Wordpress sort of standardized putting the date in the URL, I appreciate being able to spot dates in blog posts at a glance. Yes, your writing may be timeless (hopefully), but to me being able to know when something was written is important most of the time.

I think it is a good idea in some cases, but not for everything. Keeping the blog posts under the blog/ route doesn't seem like digital pollution to me.

https://dvd.qa/horse checking in!

> You can remember them.

Can you? I can remember some domains, but I can't recall many full URLs that I remember, except perhaps some that I use almost daily. Certainly not a blog post that I'll read only once.

> You can tell someone. You can even say it out loud!

If the other person is going to write down the URL, I'm not sure there's a difference here. Giving someone your email address over the phone is already a painful experience no matter what. Always have to make sure the dots are in the right place. And if the first point is wrong — "You can remember them." — then making it a little shorter over the phone isn't all that advantageous, because they're going to write it down or type it anyway, not memorize it.

The issue I see with this is that it does not allow for genericity.

URLs on sites with a ton of content often have a regularity to them. For example you might want to have different types of articles, say rooms, people and towels. For each article type you might want different affordances that usually have overlap, like an index, an input form, an activity feed...

URL paths are a good way to encode that kind of regularity.

Of course on a blog of a single person it’s whatever.

Similar: The variants of „Just write HTML!“ posts that come up every once in a while.

The "They’re enough" is ofcourse a little silly. Yes there are thousands of combinations if your short url contains any character. But his example is about words and that will leave you with less combinations.

But I like the essence of the post.

If those hierarchical directories actually work, i.e. http://example.com/blog/2022/05 actually lists blog entries written in May 2022, I'd certainly prefer that to this type of "optimization".

The only use-case I've found for short URLs is in media broadcasts (or similar). It's the only scenario I find that shortening is necessary, as the audience will have to actually type out the URL. In text formats, you generally have access to hyperlinking text to the extended URL.

I created a link shortening service that works in exactly this way, it's called pxl.to. Based on Amazon CloudFront's global edge network of 310+ Points of Presence in 47 countries. What this means for the user:

- Instant links with low latency no matter where your visitors are in the world

- Robust and reliable links that will never go down, ever

- Protection against network and application layer attacks

- Unlimited tracked clicks (no cap on clicks/month)

The service allows the creation of links using a hierarchical naming structure, as mentioned in these comments like example.com/folder/test. It also auto-generates an SSL cert for every custom domain added, for HTTPS-secure links.

I'd love any feedback! The service is largely free and can be found at https://www.pxl.to

I’m it sure what this optimises for.

Those who care about SEO use long slugs for the extra juice. Those who don’t can use IDs or whatever works for them.

Shareable URLs - still need to share the domain, and most of us (including sive.rs) don’t have a particularly memorable one.

I use mydomain.ext/long-descriptive-name for SEO but have my own url shortener, with.id/something (with.id is my short domain, and very "generic" name). Basically I follow Google like their go.gl shortener.

I would love some sort of standard for "short" URLs that are primarily designed for sharing. This would serve as an alternative to the current – usually hierarchial – URLs that are primarily designed around SEO.

I have found one truly valuable use case for shorter URLs: when a link is included in a printed document. The problem is that every URL shortener replaces the domain, which is IMO a highly dangerous thing.

I would have expected something about a super user with the 'su' url.

So short urls like this don't make sense if they are not explicit and you don't have a corresponding lookup table

I like this and I'm unreasonably happy that once I built a tiny website to share my music, where you get a bunch of listening options when visiting `example.com/listen/songname` :) another cool example would be this guys blog (look at the URL): https://there.oughta.be/a/wifi-game-boy-cartridge

I like this idea, It'll be good to have the best of both worlds. Is there a way to optimize search engine visibility for such short urls? perhaps with a good title, meta tags having keywords etc.? Is there a HTML meta tag that gives the search engine crawlers an alternate hierarchical url for the same page? I believe setting up the server to do this is not too difficult.

I create my own for this purpose;

https://xn--gda.at

It’s just a nginx with redirects on pages))

Regarding the example of instead of hi.html just name the file hi and use the nginx Config “default_type text/html” —- is there an equivalent to this for other webservers (Eg Caddy etc) ? Perhaps it would be cleaner to just create a directory called hi with and index.html?

I have a short domain name, but only as a personal URL shortener! Mine is https://m8.fyi

Yes, short is always good, I recently got http://moty.ar

Isn't remembering certain short domains hard ?

I was hearing Instagram influencers say "tap the link" all the time

so bought a domain https://tapthe.link which creates a url which can take viewers directly to the app via deep linking instead of a simple redirect,

super helpful for YouTubers and marketers and it's free for creators.

tapthelink is easy to remember than say a short url which says xlps.com or something

It's also nice to stick to the original 14-character limit on Unix file names.

don't do that

and don't do hierarchical URLs also

this are my battle-proven URL rules. whenever i violated them and changed the priority I regretted it.

  URL-rules
  URL-Rule 1: unique (1 URL == 1 resource, 1 resource == 1 URL)
  URL-Rule 2: permanent (they do not change, no dependencies to anything)
  URL-Rule 3: manageable (equals measurable, 1 logic per site section, no exceptions)
  URL-Rule 4: easily scalable logic
  URL-Rule 5: short
  URL-Rule 6: with a variation (partial) of the targeted phrase

URL-Rule 1 is more important than 1 to 6 combined, URL-Rule 2 is more important than 2 to 6 combines, … URL-Rule 5 and 6 are a trade-of. 6 is the least important.

the URL yoursite.com/short violated rule number 3. lets say you have list pages, category pages, tag pages, database generated pages, post pages .... now say you change your listpages, in design, in how they look, you want to know if they now work better (for whatever metrik you care about or not), well good luck with that. either you create a regex of hell for GA or whatever tool you are using or just a spreadsheet.

or you do example.com/l/short for list pages, example.cm/a/super-short for article pages, now easy to measure and manage.

also dont do example.com/category-name/article as all hierarchies are imperfect and change over time (there is no permanent or perfect hierarchy over time) so any kind of hierarchy violates URL rule number 2.

use %namespaces%, these are one or two letter words which identify the pagetype (listpage, article page) (note: pagetype defined as "share the same or very similar process on how the page gets created, changing the template or process of the pagetype, changes all pages under this pagetype)

i usually end up with

https://www.example.com/%pagetype namespace%/%permanent identifier i have under total control%/ or

https://www.example.com/%country or language namespace%/%pageytpe namespace%/%permanent identifier i have under total control%/ for multimarket and/or multilanguage webproperties

you can download my book for free here if you like https://gumroad.com/l/understanding-seo/hacker-news I go in a bit more detail there.

I am curious to know more about the implications for SEO.

> You can remember them. You can avoid the search engine step. No need to search when you already know the answer. Which means…

You don't know what you don't know. Discovery is about leveraging common heuristics. The more context information that's available, the easier it gets to get to the answer. Giving the least amount of information runs counter to that.

A URL is a reference, an identifier, that uniquely references a resource. A search engine essentially captures a ton of context information that you can leverage to get to a set of relevant references. An identifier can be a meaningless string of characters - e.g. a UUID - and as long as it's accompanied with context information, you can get to the resource it references.

Conversely, if you capture context in the URL itself - meaningful words, dates, authors,... - you're actually providing a breadcrumb trail for visitors - people and robots - to follow, leveraging common heuristics they might use to get to that web resource directly. Might, because discovery is always a process of making educated guesses and following cowpaths to get to the right answer.

So, no, making URL's shorter isn't necessarily advantageous.

In the same vain: long passwords using commonly, easily to remember words instead of an unintelligible string of 16 characters.

Tangentially, I also have ambiguous feelings over the widespread use of URL shorteners. Partly because they act as a middle man providing brittle URL's that can - and will - break in the long run. Partly because they hide a ton of potential context information that might be captured in the original URL.

> You can tell someone. You can even say it out loud! Whether answering an email or talking to someone on the phone, I can say, “Go to sive.rs/ff for my talk about the first follower.” or “My newest book is at sive.rs/h.” I do this often, so having memorable URLs saves me a lot of searching.

Again, context matters. This might work if you want to highlight specific content - e.g. a marketing page for your book - but it certainly doesn't work all the time for all your content. Is that blogpost from 2007 really that important that you need to be able to "say it out loud" to someone on the phone today at any moment?

Short URL's come at a cost. What's the trade off you're making here?

> They look nicer. They’re aesthetic. They show care.

I don't care. Really. I don't. I care about readability and accessibility. Sure enough, URL's with tons of non-nonsensical query parameters are a blight, but this has more to do with readability then "aesthetics". It's an URL, not poetry.

> They remove the middle-man. With long URLs, people use those ugly social share buttons that promote (and further entrench) harmful social media sites, and add visual clutter to your site. Short URLs encourage people to copy and paste the URL directly, which lets them share it anywhere, instead of only the sites for which you have a share button.

Or maybe the answer here is to avoid using social share buttons on your website at all?

> They’re enough. Using 36 characters (a-z and 0-9): 4-character URLs give you 1,679,616 (36⁴) unique combinations.You don’t need more than that.

Well, how about "sive.rs/qxfa" or "sive.rs/kxig" or "sive.rs/ddiz"? It's an argument that's in direct contradiction with the author's first argument. Again, heuristics matter. Readability matters. A big chunk of those 1.6 million odd unique combinations aren't usable of the bat because they are simply unintelligible strings of characters without meaning.

I use Restful URLs.

> Go to sive.rs/ff for my talk about the first follower

I absolutely will not. At least with known URL shorteners I know that’s what I’m getting and can inspect further. When I get links like this I assume I’m getting spam or malware. Almost all of the time it’s spam.