Hacker News Reader

Comment by AlchemistCamp | Hacker News Reader

I agree with kenji's (dead) comment.

I'd rather set up login functionality on my own server. If you use a third party service, their downtime becomes your downtime. Worse, you're injecting another company into your relationship with your users. If that were necessary, I'd choose one the customer was already familiar with, like Google, Facebook or Twitter.

Setting up login isn't very difficult and libraries exist for most webdev languages to make it even easier. It's definitely not worth 5% of my revenue.

View on HN

Login management should be built into the browsers instead. Imagine a UI where you go to log in and instead of a text/email field for login and a password field you get a drop down with a list of your personas that you’ve created with the option of adding a new one. Under the hood it would use a public/private key pair to authenticate you with the site.

Browsers could implement a service that backs up your personas and syncs them across your devices. Ideally the syncing should be cross browser too and allow me to use my own servers for this if I choose.

That sounds very much like the goals of Mozilla's Persona service. While it required a separate service and some JavaScript, the idea was that it could be supported natively by browsers in the long run.

Unfortunately, it was shut down in 2016 due to lack of interest. (Coincidentally, I just spent this past week hacking on its source code.)

I wonder if it just wasnt the right time and if the project could be revived in a couple years. It seems like a great direction.

Yes that’s pretty much what I was thinking of. It’s unfortunate Mozilla couldn’t keep it going. I wonder if a browser extension could make this work as a way to pave the way forward. Or if like LastPass could add such functionality.

I'd imagine Apple would have more success with this type of service due to the vertical integration. And they are kinda going in this direction..

https://en.wikipedia.org/wiki/Mozilla_Persona

Don't browsers essentially do this already?

You're suggesting some different backend details, but the UX you describe is already in place.

Browsers save your passwords which is not the same thing as managing your personas or using asymmetric encryption for authentication.

I don't like this idea, it makes using a public/cafe computer impossible. It would also force me to have a work browser and personal browser on all my devices. I am also not a fan of google chrome trying to log me into every site with my google profile.

It isn't necessarily any worse for public/cafe computers. Just like any login system, the UI should have the ability to remember you or not.

With the browser acting as your user agent, it can manage multiple contexts (work, personal) and let you choose which is sent to each site, without the need for separate browsers for each.

As an added tangent, consider that browsers are one of the few places where we can do things like "temporarily log me out of everything" (Incognito Mode / Private Browsing). They have some level of sophistication in session management already.

> Login management should be built into the browsers instead.

I would go even further than this, and say that a cryptocurrency might be among the best ways to build a universal login system. Imagine an Ethereum-like coin where the wallet is the username, and you spend the coins while authenticating, gaining privileges, and the blockchain stores your public info like a username and avatar that you can bring to any website.

I don’t want every login to become a part of the public ledger. Do you? Also, crypto currencies are a good solution for one thing only: electronic currencies. Outside of that they are a poorer solution compared to what’s available now. Every time.

> crypto currencies are a good solution for one thing only: electronic currencies.

What is a better solution than building a tamper proof workflow in the blockchain? Examples include:

1. I sign a contract and the notary signs and uploads a copy to the blockchain to validate the contract contents, that it was signed and by whom, and when.

2. An auditor validates a piece of evidence a client provides that proves its in compliance with a particular control in some regulatory framework. The auditor uploads a SHA256 hash of the evidence in a transaction memo to the blockchain to validate against later in peer review, a lawsuit, etc.

Edit: My examples here refer to a valid use case of a blockchain and not cryptocurrencies which is what you mention, but based on context of this discussion it looks like we all actually mean blockchain and not necessarily cryptocurrency.

Those are useful examples of a blockchain. But when I log into GMail I (a) do not want this fact to be a part of a public ledger and (b) I don’t need this to be trustless. The latter is an important point: I already trust GMail with a ton of my data including the ability to reset my credentials. There is little value gained and a lot of simplicity lost in adding a blockchain to the mix.

Blockchains have several uses but to date they are narrow. Everywhere outside of their core use cases (essentially contracts or keeping track of tokens of value) they are a hindrance. Blockchains are also hugely inefficient by design (if they were they’d be easy to crack or to perform a 51% attack on), so when you are communicating with a single third party which you trust to provide a service to you, you absolutely do not need a blockchain. We have seen time and time again people trying to throw this solution at all the problems from DNS to buying groceries and yet the only solutions still worth talking about remain cryptocurrencies. Turns out well defined client/server APIs and databases actually do work about as well as anything else.

In the spirit of discussion, what could you do if you involve a blockchain in logging into GMail that you absolutely cannot do without it? If the answer is nothing, then there is zero reason to go down that path.

I agree with your points and overall sentiment, I'm simply rebutting the initial statement in your parent comment,

> crypto currencies are a good solution for one thing only: electronic currencies.

You contradict yourself by stating above you believe there's only one valid use case for blockchain, but in your most recent comment you obviously admit there are several valid use cases (though admittedly a limited number). Maybe your initial comment was meant for emphasis and not literally that you believe there is only one use case, but be careful using quantifiable verbiage or hard numbers to portray emphasis without explicitly stating that's what you're doing, your credibility gets called into question when doing so.

I definitely don't think it's /useless/, but I wouldn't use it. Someone out there wants a dead simple login solution that they can get running with a handful of lines of code.

That being said, if this was /self hostable/, I'd see myself using it on a few projects.

Off-topic side remark

HN ‘markup’ diversity is always a joy for the punctuation mark enthusiast.

\Hooray for markup/

(You know, you can get proper italics. Wrap it in asterisks. Some discover this by accident when it mangles asterisks they meant as literal asterisks.)

I used it that way because I've gotten used to the org-mode syntax. It's much easier to /me/.

While there often are built in modules for lots of platforms, it often becomes not so easy out of the box when a platform scales. Adding login to RoR is easy, but what if there suddenly is a wordpress site serving the articles, some java backend services handling other aspects etc. Or just generally lots of services that are separate. Then one suddenly needs a central authority not built in to any one service.

Then I've seen lots of people offload that to Auth0 or other managed services.

As an Auth0 user, I agree it’s not hard to setup user logins for most frameworks. The reason I’m using Auth0 boils down to business reasons not technical ones.

- If I cant make enough money per user to even pay for Auth0, why am I working on this?

- Auth0 provides more than login, its got logic related to token re-validation, signup pipelines, social platform integration, etc... sure, I could build each of these as I needed them, in fact I did previously build them for myself. However by using Auth0 I know in advance that all this kind of functionality will be available when/if I need it.

- Easy cross system auth. If I want to build something using separate services, like a front end in one tech stack and a backend/admin using a completely different stack, Auth0 makes it so easy to share logins, roles, and permissions between the two systems without having to work out how to handle it myself, or within having to run my own third system like my own installation of Keycloak or building my own equivalent to it.

That would be an ideal scenario for a SSO system. There are open source solutions like Keycloak, IdentityServer etc. While they are nice thanks to the functionality they offer, I recently tried Keycloak and found it extremely annoying to maintain thanks to the not great documentation.

Managed services like Auth0 are a great alternative, although Auth0 is one of the more expensive ones. For small apps I found Firebase Auth to be great, for larger projects things like Amazon Cognito or Azure AD B2C are probably the better choice (although Azure AD B2C still doesn't support custom domains). They are worth a try imho, since Amazon as well as Microsoft offer a generous free teir (50k MAUs for free)

I think most people here would rather build it themselves but there are always people out there who would pay for someone else to do it.

Reply on news.ycombinator.com