I seriously thought they had ported some novel browser engine to Win16, but even that, Windows 1.0, which IIRC didn't even have overlapping windows, would be ridiculous.
And apparently it's based on Chromium. Porting a brand-new browser engine to long-obsolete Windows is probably plausible. Porting Chromium, a very large beast with a lot of baggage, is much less likely.
Arc on Windows still feels like the early versions of Arc on macOS, but I’ve genuinely enjoyed my beta access anyway. I also love how transparent the browser company is.
I hope the company can find some way to stay alive, but I’m not sure what a path to profitability would look like for them. Subscription fees?
Although it's easy to dismiss as Yet Another Chromium Fork there is some technical novelty here, because their custom frontend is probably the first serious Swift project outside of Apples ecosystem. Maybe Swift will finally get a foothold on Windows?
Just getting HTML working would be mind-blowing on a Windows 1.0 machine, assuming you were running the typical hardware of the time.
From what I can tell, the 80286 was still a server-grade CPU in 1985, and most PCs were still running an 8086 or 8088, which maxed out at 1 MB of RAM. Just the HTML for the Wikipedia article on the 8086 [0] is nearly 224 kilobytes, nearly 1/4 of the RAM available.
I'd think it could do TLS encryption, it just wouldn't do it FAST. It would also probably be limited in the ciphers and key exchange algorithms it would support.
While you’re technically correct, I don’t think it captures the spirit of my comment.
Modern (in comparison) CPUs contain instructions to aid with performance, such as AES extensions. If you’re doing everything in software then you’ll be limited to the older ciphers, most of which might not even be supported by modern sites given TLS1.3 is the recommendation. For example OpenSSL doesn’t support RC and AES will benchmark much slower on CPUs (when compared with RC) without AES instructions and smaller L1 cache.
Sure you could workaround that by supplementing cache with writing to persistent storage and just throwing more time at the problem. But then you’re looking at days or even weeks just to complete the handshake, let alone pulling an HTML content. By which point the novelty of getting an 8088 online would long since have worn off.
So to say TLS wouldn’t be fast is an understatement. Calling it an understatement is itself an understatement :)
If you remove TLS from the equation things get dramatically simpler. I’ve got a 64k 8bit micro (Amstrad CPC 464) with a WiFi adapter. The adapter handles the wireless protocols but beyond that I wrote a very simple HTML browser. Though most other similar projects I’ve seen have used a Raspberry Pi hooked up via serial to provide offloading. But that’s basically turning your 8bit Micro into a dumb terminal and thus feels somewhat like cheating.
> For example OpenSSL doesn’t support RC and AES will benchmark much slower on CPUs (when compared with RC) without AES instructions and smaller L1 cache.
> Sure you could workaround that by supplementing cache with writing to persistent storage and just throwing more time at the problem.
I think my question is...if you only supported a single key exchange algorithm and AES-128, and didn't verify server certificates, how much code would a TLS negotiation take? My understanding is that TLS libraries get large because they support dozens of key exchanges and ciphers, plus support previous versions of TLS, and a huge suite of features that most people probably don't even use. I would THINK you could achieve a bare minimum that an 8088 could run in under 100K of code.
Damn, now I'm tempted to actually try this. Know any way to emulate an 8088 at an accurate speed? ;-)
The problem isn’t code size for unused cyphers. The very first part of any SSH or TLS handshake is agreeing to what cypher suit to communicate on.
The problem is purely the computational overhead of the encryption itself. This is why modern CPUs have instructions to offload some of that overhead to hardware.
You can think of this as the same kind of problem of MPEG encoding and decoding in software vs hardware. The difference in performance is massive. Then try to do that on a 40 year old 16 bit CPU with virtually no cache and only 1 MB of RAM to play with
I used Flock but it was no better than FireFox - the skin/theming was nice though.
It was another VC funded project that nowhere when it came how do you actually make money , that said I hope them the best - Firefox showed us the way to building a better web browser so maybe who knows.
It was definitely fun when it worked! It's a shame Firefox (well, Gecko) isn't nearly as repackageable/embeddable as it used to be. Alternative gecko browsers like Camino were so good.
Yeah the initial GA is for Windows 11 only.
I'm also curious about the Windows 10 support release date.
According to their newsletter, they plan to answer this question on their next YouTube video.
Technically you can force your way to install Arc on Windows 10[1], but it was a rough experience when I tried few months ago during beta.
Windows 11 comes installed on most people’s computers, and among people on this site I’m pretty sure most of us have either had to update for some reason or just auto-updated one day. I don’t know anyone who is skipping Windows 11, given that it takes nontrivial effort to do so.
Conversely, I only know one person on Windows 11, and he just asked another friend for help rolling back to 10 due to how miserable support for his most used applications are.
Much like Windows 8, most people on Windows 11 tend to be the technically-disinclined, which generally are not the userbase of an experimental alternative browser. Thus I too, am understandably baffled at the lack of Windows 10 support.
A lot of people still dislike Windows 11, in no small part because it released in an unfinished state. I think 23H2 has finally made Windows 11 something work upgrading to, but if you're used to Windows 10, you'd never know that
"it takes nontrivial effort to do so" - not sure I agree. My desktop (high-ish spec when I bought it in 2020) came without TPM 2.0 enabled, which effectively blocks 11. I could probably enable it, but that to me looks like the nontrivial effort.
Windows 11 requires significantly more advanced hardware and power to even be installed. Meanwhile, you could install Windows 10 on an WinXP-era netbook and it would be functional, if laggy on 2GB RAM.
It doesn't seem to require anything more, except maybe the whole TPM 2.0 thing.
Who is running 2GB of ram these days? Win 11 recommends 4GB, and imo that's very little on a modern computer. 8GB for casual use, 16 for anything more intensive.
Needs parens. They mean "Arc Browser [for Windows] 1.0, Now generally available". However, I got excited since I read that as "Arc Browser (for Windows 1.0), Now generally available" which would have been actually cool.
I looked it up, it's an official word - at least on Wikipedia:
"Enshittification is the pattern of decreasing quality observed in online services and products such as Amazon, Facebook, Google Search, Twitter, Bandcamp, Reddit, Uber, and Unity. The term was used by writer Cory Doctorow in November 2022, and the American Dialect Society selected it as its 2023 Word of the Year. Doctorow has also used the term platform decay to describe the same concept."
- https://en.wikipedia.org/wiki/Enshittification
While Arc doesn't have subscription model yet, I can see them going in that direction.
From their FAQ[1]:
> To do that, we’re excited about business models that align our incentives with the people who we serve: from charging companies that want to increase the productivity of their teams across the many tools they use for work, to making it easier and safer to pay for things online.
So far they have been doing various small QoL for popular websites[2].
My favorite one is GitHub Live Folders[3], which is kinda like automatic bookmark/pinned tabs for your own GH PRs.
I can see people paying for these collections of small nice things if they keep adding more.
Of course they also AI-powered features, dubbed Arc Max[4].
Personally I often use the page summarizer, which feels like lite version of Kagi Summarizer.
Right now seems like they're burning money by running it for free.
But they could opt to charge money, e.g. for more powerful model.
Disclaimer: I'm just a happy Arc user, it's not perfect, but I use it daily side by side with Firefox.
Their interface on top of Chromium is written in Swift, so they're bundling all the stuff that is necessary to make Swift run on Windows, top of of Chromium.
Started using Arc on windows a few months ago. Everyone I have recommended it to loves it. It's hard to say exactly why. You can duplicate a lot of the functionality with tree style tabs and other extensions but for a work browser it's been a dream.
I started using Arc ~1 year ago. I like it's model of tab management a lot, it's made using the browser much easier. Around ~2 months ago I thought maybe it was running slow and went back to Edge. Well, never really got a clear sense of if Arc was slow, but I missed it's tab management features so much, I had to switch back.
- Favorites are "sticky" to the original URL you saved them under. So you can navigate within that tab, but you can always click to get back to the original favorited URL.
- Regular tabs are auto-archived if not used in 24 hours.
And some other features:
- Vertical tab bar
- Command palette for searching/opening new tabs
- Folder structure for organizing favorite tabs.
- "Spaces" (like virtual desktops) which tbh I don't see much use from.
- Split tabs
- Some AI features they are pushing now. Don't really use these much either.
- Lots of integrations, eg, hover to see your GCal. They just pushed a new "live folder" feature that auto organizes all your open GH PRs. This is super useful.
All in all, I might actually pay for this if they make it subscription only. For me it's the difference between having 150 tabs open in Edge (many of them duplicates or one-offs) or 20-30 tabs open in Arc, and never thinking too much about tab management. All in all I think it points to the fact that the browser space is sorely under-optimized for power users. The #1 tool everyone uses day in and day out and most people just accept having 200 unmanageable chrome tabs open as the cost of doing business?
Spaces are great in combination with profiles. I have a space/profile for work and personal, since my work requires a few different extensions, Google login, etc
It is good to see that they are still requiring you tell them your email address to try the browser. Too many browsers attempt to gain “market share” by dirty tricks like “being easily available” and losing focus on the obvious #1 priority, amassing the largest possible email list
Agree 100%. I only hope that they start requiring users input their mailing addresses and social security numbers in order to download the browser. Then it'd be a no-brainer.
I like this idea. Why shouldn’t The Browser Company know where I live? They could mail me important browser update news, which is perfectly logical because it is imperative that The Browser Company has a way to contact me directly
I do not understand how a browser can launch in 2024 without caring for privacy. Arc is probably the worst browser ever in this regard, making it mandatory to have an account.
Brave would grow more if it stopped stuffing in every crypto blockchain feature it possibly could. I swear, every 3rd update adds some new annoying feature to turn off.
I've been daily driving Brave on every machine and device I have for a year or so now... and I really wish they would have just stuck with the "Privacy Focused Chromium" thing. They've wandered far down this path of "lets just add every feature that comes to mind" and it's really not helping it's usefulness.
That's because the competing browsers are owned by some of the largest corps on earth that invest a lot in to making sure their browser has as much share as possible.
You may as well say that moviegoers don't care about art because studios only finance superhero franchise films.
I've only used Arc a bit, and generally had a good experience, but I do wish that they had chosen not to rely Chromium. I understand that would have massively increased the amount of work required initially, but I really worry about our collective reliance on Chromium. I don't know... maybe I'm a pessimist.
I’d recommend people give it a try - if only to see what some properly different thinking looks like about what a browser could be. There are some really interesting UI things going on in Arc, and if all it does is makes you go “oo, interesting” then it’s worth it. There are so many boring paradigms repeated all the time, particularly in browser ui’s that it’s worth a shot.
And to anyone bemoaning the “must have account” thing - this is HN, I’m sure you’ve all got the technical nouse to set up a burner account if you’re that uncomfortable. Or SimpleLogin.io - another HN favourite ;-)
If you can't even use it without an account, then that's an issue. It's not like Kagi where using the tool requires compute credits on their side, so offering it without a login wall isn't possible.
Mozilla wants me to sign up for a Firefox account too. The difference is that it's entirely optional and the browser works just fine without it.
It is, because other browsers don't require it. I would not have tried out Chrome in '08 if I needed to log into it. What's next? 2FA to secure access to the thing so I can 2FA all the other things I have logins for?
Besides giving them an open invitation to spam your email, login based access tells me that they're doing what Chrome is, attaching your browser history to your identity. Google does this because they need a way to fuel their behavioural ad machine without cookies. I imagine Arc is doing it for the same reason - monetization.
There seem to be a lot of PR phonies in the comments area. I tried it out and found that the UI is quite confusing and difficult to use. It looks absurd due to its non-customizable design. Rounded corners in your page view, really? I removed it instantly. Since Firefox and Edge have split view, vertical tabs, and profile synchronization much ahead of this browser, I'm not sure who it's meant for.
It’s the principle of the mandatory account which really bugs me, more than the technical logistics of it.
Not only does it seem unnecessary to require this even just for curious new users to quickly try it out, but it sends me a strong signal about their priorities and tells me that they probably don’t align with mine.
I'm not sure if simply having an attractive UI is enough to disrupt the browser industry. For Arc to make a dent in the desktop browser market, they need game-changing features like the 'Browse for Me' available on desktop
Installed it, fancy onboarding for gen z and all that, but wow, a wall of mandatory Arc account creation to start using the browser. At least they require * only * email.
Thanks that email mask services like Firefox Relay exists, I'm using it for everything lately, with the push of surveillance from tech companies and data breaches I feel a bit safer at least.
I was surprised to know that it's only for Windows 11. I can't remember a program that launched without supporting the previous version of Windows. It seems to have a workaround to install it on Windows 10, but I guess I don't want to try it that bad.
95 comments
[ 2.9 ms ] story [ 105 ms ] threadhttps://arc.net/
And apparently it's based on Chromium. Porting a brand-new browser engine to long-obsolete Windows is probably plausible. Porting Chromium, a very large beast with a lot of baggage, is much less likely.
that's why arc supports split browser tabs hahaha
It's a desktop environment for DOS that was released in 1985 :)
Arc on Windows still feels like the early versions of Arc on macOS, but I’ve genuinely enjoyed my beta access anyway. I also love how transparent the browser company is.
I hope the company can find some way to stay alive, but I’m not sure what a path to profitability would look like for them. Subscription fees?
My initial reaction was “oh that’s seriously cool. I wonder what TCP/IP stack they’re using and if they got CSS working or just HTML”
From what I can tell, the 80286 was still a server-grade CPU in 1985, and most PCs were still running an 8086 or 8088, which maxed out at 1 MB of RAM. Just the HTML for the Wikipedia article on the 8086 [0] is nearly 224 kilobytes, nearly 1/4 of the RAM available.
https://en.wikipedia.org/wiki/Intel_8086
It definitely wouldn’t be powerful enough to do TLS encryption either.
I'd think it could do TLS encryption, it just wouldn't do it FAST. It would also probably be limited in the ciphers and key exchange algorithms it would support.
Modern (in comparison) CPUs contain instructions to aid with performance, such as AES extensions. If you’re doing everything in software then you’ll be limited to the older ciphers, most of which might not even be supported by modern sites given TLS1.3 is the recommendation. For example OpenSSL doesn’t support RC and AES will benchmark much slower on CPUs (when compared with RC) without AES instructions and smaller L1 cache.
Sure you could workaround that by supplementing cache with writing to persistent storage and just throwing more time at the problem. But then you’re looking at days or even weeks just to complete the handshake, let alone pulling an HTML content. By which point the novelty of getting an 8088 online would long since have worn off.
So to say TLS wouldn’t be fast is an understatement. Calling it an understatement is itself an understatement :)
If you remove TLS from the equation things get dramatically simpler. I’ve got a 64k 8bit micro (Amstrad CPC 464) with a WiFi adapter. The adapter handles the wireless protocols but beyond that I wrote a very simple HTML browser. Though most other similar projects I’ve seen have used a Raspberry Pi hooked up via serial to provide offloading. But that’s basically turning your 8bit Micro into a dumb terminal and thus feels somewhat like cheating.
> Sure you could workaround that by supplementing cache with writing to persistent storage and just throwing more time at the problem.
I think my question is...if you only supported a single key exchange algorithm and AES-128, and didn't verify server certificates, how much code would a TLS negotiation take? My understanding is that TLS libraries get large because they support dozens of key exchanges and ciphers, plus support previous versions of TLS, and a huge suite of features that most people probably don't even use. I would THINK you could achieve a bare minimum that an 8088 could run in under 100K of code.
Damn, now I'm tempted to actually try this. Know any way to emulate an 8088 at an accurate speed? ;-)
The problem is purely the computational overhead of the encryption itself. This is why modern CPUs have instructions to offload some of that overhead to hardware.
You can think of this as the same kind of problem of MPEG encoding and decoding in software vs hardware. The difference in performance is massive. Then try to do that on a 40 year old 16 bit CPU with virtually no cache and only 1 MB of RAM to play with
Technically you can force your way to install Arc on Windows 10[1], but it was a rough experience when I tried few months ago during beta.
[1]: Guide to installing Arc Browser on Windows 10 - https://gist.github.com/TrevTV/2044e43666a8fa4bb581d4b0c8316...
Much like Windows 8, most people on Windows 11 tend to be the technically-disinclined, which generally are not the userbase of an experimental alternative browser. Thus I too, am understandably baffled at the lack of Windows 10 support.
https://learn.microsoft.com/en-us/windows/dev-drive/
Unlike Vista and 8, nothing broke and everything is where I expect, so I've stuck with it.
Who is running 2GB of ram these days? Win 11 recommends 4GB, and imo that's very little on a modern computer. 8GB for casual use, 16 for anything more intensive.
"Enshittification is the pattern of decreasing quality observed in online services and products such as Amazon, Facebook, Google Search, Twitter, Bandcamp, Reddit, Uber, and Unity. The term was used by writer Cory Doctorow in November 2022, and the American Dialect Society selected it as its 2023 Word of the Year. Doctorow has also used the term platform decay to describe the same concept." - https://en.wikipedia.org/wiki/Enshittification
[1] https://americandialect.org/2023-word-of-the-year-is-enshitt...
What's an unofficial word?
From their FAQ[1]:
> To do that, we’re excited about business models that align our incentives with the people who we serve: from charging companies that want to increase the productivity of their teams across the many tools they use for work, to making it easier and safer to pay for things online.
So far they have been doing various small QoL for popular websites[2]. My favorite one is GitHub Live Folders[3], which is kinda like automatic bookmark/pinned tabs for your own GH PRs. I can see people paying for these collections of small nice things if they keep adding more.
Of course they also AI-powered features, dubbed Arc Max[4]. Personally I often use the page summarizer, which feels like lite version of Kagi Summarizer. Right now seems like they're burning money by running it for free. But they could opt to charge money, e.g. for more powerful model.
Disclaimer: I'm just a happy Arc user, it's not perfect, but I use it daily side by side with Firefox.
[1]: https://arc.net/faq
[2]: https://arc.net/integrations
[3]: https://resources.arc.net/hc/en-us/articles/22731612065815-A...
[4]: https://arc.net/max
Last time i checked they had something like 60 employees (in New York?). Burning money indeed.
But why is the Arc Browser package size >700 MB? It’s double the size of Chromium! Where is all that bloat coming from?
The submission was earlier titled: Arc Browser for Windows 1.0, now generally available which is confusing as you can see in the comment thread.
It's the 1.0 release of Arc Browser for Windows.
To summarize the Arc Tab Management Philosophy:
- 3 Groups: Pinned Tabs, Favorites Tabs, "Regular" Tabs
- Pinned tabs never go away
- Favorites are "sticky" to the original URL you saved them under. So you can navigate within that tab, but you can always click to get back to the original favorited URL.
- Regular tabs are auto-archived if not used in 24 hours.
And some other features:
- Vertical tab bar
- Command palette for searching/opening new tabs
- Folder structure for organizing favorite tabs.
- "Spaces" (like virtual desktops) which tbh I don't see much use from.
- Split tabs
- Some AI features they are pushing now. Don't really use these much either.
- Lots of integrations, eg, hover to see your GCal. They just pushed a new "live folder" feature that auto organizes all your open GH PRs. This is super useful.
All in all, I might actually pay for this if they make it subscription only. For me it's the difference between having 150 tabs open in Edge (many of them duplicates or one-offs) or 20-30 tabs open in Arc, and never thinking too much about tab management. All in all I think it points to the fact that the browser space is sorely under-optimized for power users. The #1 tool everyone uses day in and day out and most people just accept having 200 unmanageable chrome tabs open as the cost of doing business?
Brave has <1% market share.
I've been daily driving Brave on every machine and device I have for a year or so now... and I really wish they would have just stuck with the "Privacy Focused Chromium" thing. They've wandered far down this path of "lets just add every feature that comes to mind" and it's really not helping it's usefulness.
You may as well say that moviegoers don't care about art because studios only finance superhero franchise films.
Do they? Do you think the audience drives the genre of movies that are popular or vice versa?
We have Firefox and Safari still though so it is not quite that dark yet but we are getting there.
Making a browser these days from scratch, or even keeping up with the rest is incredibly time consuming and expensive.
https://github.com/thebrowsercompany/swift-winrt
https://github.com/microsoft/cppwinrt
https://github.com/microsoft/windows-rs
And to anyone bemoaning the “must have account” thing - this is HN, I’m sure you’ve all got the technical nouse to set up a burner account if you’re that uncomfortable. Or SimpleLogin.io - another HN favourite ;-)
Mozilla wants me to sign up for a Firefox account too. The difference is that it's entirely optional and the browser works just fine without it.
Besides giving them an open invitation to spam your email, login based access tells me that they're doing what Chrome is, attaching your browser history to your identity. Google does this because they need a way to fuel their behavioural ad machine without cookies. I imagine Arc is doing it for the same reason - monetization.
Not only does it seem unnecessary to require this even just for curious new users to quickly try it out, but it sends me a strong signal about their priorities and tells me that they probably don’t align with mine.
Thanks that email mask services like Firefox Relay exists, I'm using it for everything lately, with the push of surveillance from tech companies and data breaches I feel a bit safer at least.