31
votes
The Browser Company announces Arc Browser will no longer be their flagship product
Link information
This data is scraped automatically and may be incorrect.
- Title
- Josh Miller on X (formerly Twitter): "Grateful for the outpouring the past few days. I can't tell you how motivating it is to have people care so much about @arcinternet. The encouragement, the criticism, the confusion. Took it all in <3As a thank you, I'd like to speak more plainly about what's happening and why -... pic.twitter.com/g3i0uSFj4t / X"
- Authors
- Josh Miller
In my opinion, the Browser Company has completely lost the plot. The only reason they were so talked about when they came up was that Arc was sort of the only browser that did what it did. Before the GPT revolution, mind you. The idea was that it had easily sortable vertical tabs and viewed browser windows as static, not disposable. They wanted you to have one browser window, not however many you wanted. To some, especially to those who work mostly in web apps online - think Google Docs, Slack, Office, Airtable, Notion and all the other ones - this made sense. In fact, there was an app called Station a few years back that did the exact same thing, but I don't know or care to check if they're still around.
Then the GPT revolution came, and everyone wanted to be first to introduce Al features. Not many of them succeeded, and I think I can confidently say that we're sick of this corporate dick-measuring by now, and Arc was no different. Arc Max, their free addition to the Arc browser, included things like summarising web pages, renaming downloaded files and open tabs, and an "enhanced" Cmd+F that doesn't just search for text in pages, but answers questions. Having used Arc for more than two years before switching to something else, I can say that those never worked properly.
Their product stack sort of broke as a result. People wanted Arc on mobile, and on windows - not particularly because it worked for their workflow, although it did for some of them - but mainly because it was the shiny new thing with the cool design and all the hipster cred. Most people don't even use the stand-out features that made Arc special, like Easels, Notes and Split Screen. They were just sick of Chrome.
Course, Arc is just Chrome with dipping mustards. And when the Arc Companion for iOS released, not many people used it. It was only when they announced Arc Search, a seperate app (So we're at three now, well before this annoucement) that people started to care. And guess what: Arc Search is Arc in name only. It doesn't do any of the productivity features like Easels or Tab grouping, it's a wrapper for yet another Al-powered chatbot search, in a pleasing visual style. That's it.
So when they say that Arc isn't their flagship endeavour anymore, I can't really say I'm surprised. I think if you asked the Browser Company directly and asked what Arc is supposed to be, right now, they would have a hard time answering. Should it make browsing easier through Al slop? Should it make productivity easier through the features it shipped with? Hard to say. It's unfocused now, and that's a real shame because there are obviously skilled and talented engineers and designers at BCNY.
I guess I don't have much hope for the "next Arc", or whatever it will be called. Arc wasn't good. Arc can't be good, in part because it doesn't know what it wants to be but also in part because it can't be something greater than the sum of its barely functional parts.
Use Orion, or Firefox. Don't bother with this.
When Arc got announced and it was discussed here on Tildes, I wrote a comment detailing my reservations about adopting Arc as my main browser. It's very clear that the company making it is following the typical tech company startup scheme, i.e. get seed funding, grow fast, worry about how to monetize it later.
You can read between the lines of the CEO's announcement video that this is exactly what's happening. Their key performance indicator is user growth, and the CEO said in the video that with the current growth rate, they are not going to reach a billion users. Read, or listen, in this case, between the lines: they are not growing fast enough.
Browsers have been free for over two decades. Chrome exists because Google wants data, Firefox exists because Google really doesn't want Chrome to be designated a monopoly, Safari exists because Apple is obsessed with end-to-end control. And Browsers are the only thing this company is presumably going to be making.
They have no path to profitability, they don't even have the concept of a plan of profitability, and they are not growing fast enough to make the people funding them right now happy. This is what this is. They made something good, but too niche for tech investors. They are also headquartered in New York, so I imagine their upkeep is expensive.
This is what this is. They cannot, or are unwilling for whatever reason, to monetize their current userbase. So to get more funding, they have to keep growing. This is their attempt.
It's tangential and perhaps nitpicky, but I think Safari actually came into existence because at that point in time, the Mac browser landscape was underwhelming.
The options were mainly Internet Explorer which was decent (considerably better than IE 5/6 on Windows in fact) but had been put into maintanence mode by MS and Netscape/Mozilla Suite which was a clunky beast. Firefox (known as Phoenix and Firebird) was still in alpha/beta stages and an unknown quantity. Mac users at that point (myself included) tended to jump between browsers depending on what they needed and how much they could tolerate Mozilla Suite's sluggishness and instability. Not a great picture for browsers, which even then had become one of the most important programs for desktop platforms.
So Apple took things into their own hands by forking KHTML into WebKit and building Safari around it.
It makes sense to have a native browser for your operating system. I'm not denouncing Safari, but what I mean is that for Apple, Safari is part of a package that sells their hardware, so it can be free in the sense that you need to buy a device from them to get to it. They don't need to monetize Safari itself.
The Browser Company however, does. It's in their name, they want to make Browsers only. But clearly they are still in the stage of achieving infinite growth so this is what they do. Don't build a stable product and focus on serving the user base (which, if they didn't have investors hounding them, they could at least attempt) instead of, I'm assuming here, putting Arc in maintenance mode and developing something else. It feels to me like Arc is being put into maintenance mode.
I would say that part of Arc’s appeal is also in its visual design, animations, and general impression of polish.
None of the other cross platform browsers really come close. Chrome has been plain ugly for a while now, Firefox looks unremarkable and frequently has rough edges that appear here and there momentarily. Edge has the same problem as Chrome with a tinge of MS flavor added and paradoxically feels out of place on Windows — somehow Arc is the most native-feeling browser there despite having started Mac-only!
Right now my main browser is Orion, which has improved immensely in the past few months to a year but still hasn’t quite reached Arc’s level of polish yet. The dev is highly responsive to feature suggestions and bug reports though which is nice. It only runs on macOS though, so on Windows and Linux I’m using Firefox with the new experimental vertical tabs enabled.
Check out the Zen browser. It’s basically Arc on Firefox without the bullshit. It also has the designer/UX/animation appeal.
I've played with Zen and it's promising, but there's a certain… clunk factor that everything based on Mozilla's tech stack has and it remains to be seen if the Zen team can overcome that.
That was why I used arc. Modern browsers have so much cruft because of legacy decisions. There are a ton of UI decisions that are obviously wrong or bad today, but stay because they were made a long time ago. Nobody in recent years has designed a browser from the ground up for modern browser usage. Arc did. That is why I liked arc. It was pleasant to use, and I enjoyed browsing the web when I used arc.
But I am very willing to work around software inadequacies when it is actively being improved. I am much less willing to work around bad software when it will never be changed. So now I am back on safari. It’s not as pleasant, but it does some things way better and will actually be improved over time.
In this case I strongly recommend Zen Browser. It's a Firefox fork with overhauled UI centered around vertical tabs and a compact mode that saves as much screenspace as possible.
It also has split tabs, pinned tabs, workspaces, glance (just introduced) and development is currently very active.
Has Orion improved on iOS? I tried it 6 months or so ago and grew frustrated quickly.
Also, haven't loved the founder's ethics. Partnering with Brave/Eich.
It's still rough around the edges, but it is the only browser insane enough to allow Firefox and Chrome extensions simultaneously, on iOS. That's unheard of, especially because it uses neither of those as web renderers, and it does make at least iPad browsing much more desktop-class.
When it comes to the Brave partnership, they walked that back in January, as far as I can tell. They use Brave's API sometimes, for what they claim is around 10% of search queries, but there's no exchange of money to Eich or his company.
I haven’t used the mobile version in a long time, but last I did it definitely trailed behind the desktop version in quality. Desktop is the focal point.
On ethics, I can’t say I’m without reservations but there are so few options, especially if you want native vertical tabs and something that tries to fit the platform instead of doing the “UI as branding” thing that everything except maybe Safari does.
The latter of those is particularly frustrating to me; I don’t want my browser to be distinct, it’s just one among many tools on my computer/phone, just like utilities like calculators and text editors and should act accordingly. There’s no good reason to repeat all the work on usability, accessibility, etc done by host OSes — it’s a costly distraction and waste of energy that could’ve been put towards tangible improvements in features, efficiency, etc.
The original tweet is so vague, I’m not sure what they’re trying to say that couldn’t have been written as “our browser idea didn’t take off like we hoped, so we’re working on something else”.
TBH I found their whole proposition kind of weird.
“Sign up to some random company to use a browser that’s just…another chrome clone, but now with vertical tabs and more AI”?? Like…why?
Archive: https://archive.is/y3VWd
More context:
It appears this tweet is a direct follow-up to their recent video titled "What have we been up to? (CEO Update)"
This major announcement was at the bottom of their recent release notes
Commentary:
I've been using Arc for a year and a half now. While I noticed their slow-down in the past few months, I didn't expect this outcome. It sounds like it's been placed on long-term support with no new features planned. I've reported a handful of bugs to them since switching. Two medium-priority issues still haven't been fixed for months, despite direct responses from staff members stating they were working on a fix. It seems their staffing has been heavily reallocated.
It's pretty disappointing to see how much the company is doubling down on LLMs to replace user agency on the web. I suppose I'm glad the original Arc won't be replaced by a very different 2.0 version. However, I've still lost a lot of confidence in their ability to manage Arc effectively while on the back burner.
I set up ArcWTF recently since I found the official Arc for Windows application to be lacking. So maybe I'll try that on other devices. I was an avid Firefox user before Arc, but with the recent Google monopoly ruling I fear for their future as well. TechAltar's recent video does a good job of explaining why Mozilla appears to be pivoting into advertising since they're not financially independent.
Maybe I'll try looking at Zen Browser again. I spend so much time in a browser that changing feels like a major life decision.
I was an Arc early adopter, but I've since moved back to Firefox. I could see the writing on the wall several months back when the release notes essentially became marketing releases. I also didn't want to stay on a Chromium based browser any longer due to the manifest v3 fiasco. I tried out Zen, but other than a similar-ish UI to Arc, it didn't offer enough over plain old Firefox.
Honestly I've never even heard of Arc, and the fact that it's proprietary would have kept me from trying it out anyway. Because no matter how good it is or was, a free web browser released by a for-profit company will eventually need to do some anti-consumer shit to become profitable, and there'd be not a damn thing you can do about it.
I've been Firefox-only for years. I'm not that interested in a Chromium-based browser which is chained to Google's vision for the web. The only reason I'm even considering looking for alternatives now is that Mozilla Foundation has laid off a third of their staff, and I'm worried they're past the point of no return. If they're that worried, why haven't I heard any appeals for donations, etc? They're just kinda quietly corporatizing and finding a path to profitability, which will surely damage their social mission.
The good news is that Firefox is open-source and something new can rise from the ashes of Mozilla if/when they drive themselves into the ground. Chromium is open-source, too, but you'll never be able to fork it and cut your own path as long as it's in Google's clutches... and seeing that Google is one of the biggest companies in the world, and controlling how you access the web is one of the things that keeps them in that position, I don't see that changing any time soon.
I can vouch for Zen. I've been using it for the past few months and I'm enjoying it. Of course, if you don't like Vertical Tabs, then Zen isn't for you.
Keep in mind it's very alpha/active development, so things keep changing and everynow and then there is a major bug, though the maintainer always fixes those major bugs ASAP.
Arc's "I'll summarize web pages for you" feature was the initial draw for me. But then I came to learn that:
Sure, LLMs are improving quickly. And yet some things remain the same. They still:
In a nutshell, their own biases often seem to work contrary to the user. I haven't seen this changing rapidly. Perhaps not much at all.
Tangentially, LLMs are also still an absurd burden on the environment. Their rate of energy consumption could be a dictionary definition example for "profligacy". Meanwhile, they, still, are the new gold rush. That gold rush threatens to increase the acceleration of Earth's de-teraforming.
LLM’s are pattern-matching engines and an easy way to get it to follow a pattern (which is what it wants to do) is to give it a few examples and ask it to continue the list. I wouldn’t say this is user-hostile so much as defying unrealistic expectations that they’re magic. The wide-open UI (ask any question) provides little guidance about what they’re good for, compared to more traditional UI.
One argument for running LLM’s on your own machine is that you know how much energy you’re using. If it’s running down the battery, you’ll notice. The ones that are practical to run locally don’t use as much energy. They’re also less impressive, but they’re improving.
You’re right about the gold rush, but we can hope things get better as efficiency improves, people get a better idea of what they’re good for, and available products become more taylored to specific use cases.
How does LLM energy consumption look when broken down into training costs and ongoing per-query costs?
Running queries (called inference) is significantly cheaper than training the models. It's still not free, but then again neither are things like Google searches either.
There's a lot of research right now into reducing the resource requirements of both training and inference. Some examples are exploring alternative and hybrid architectures like Mamba, lower bit models like BitNet b1.58, caching methods for repetitive prompts, and various quantization strategies (GGUF, GPTQ, EXL2, AWQ, et al).
It's a big problem space, and there's a lot of research into different approaches to make it more efficient and reliable.
edit: Fixed a loose thought.
The founder of Kagi, the search engine, has previously stated that LLM features cost them approximately 1/10th of a web query, which really surprised me.
While I knew searches need quite a bit of energy in comparison to people taking them as granted for free/not thinking about it, still, I would’ve expected for LLMs to be much more costly on average.
Also, I’m not sure whether this holds true across the industry or just for them.
Edit: grammar/typo
Likely many factors to this, but looking at kagi.com/stats the majority of users uses Kagi for search, which comes with close to a fixed of around $0.015 (at least it was around that last time I saw the topic on the Discord). While Kagi certainly have some LLM power users, most likely only uses it occasionally. It is comparing apples and oranges, but gpt-4o costs $0.00250 for 1000 tokens (not really sure how much that actually is). I would also guess that currently that price is set too low, since the LLM industry is heavily subsidized by investors, hiding the actual costs.
This is something I almost touched on in my comment as well, and I agree. If for example OpenAI had to offer e.g. GPT models at-cost instead of operating at their current loss, they’d be in huge trouble (moreso than they are already).
Good question! Beyond my knowledge to answer. Curious though.
I expect training will be ongoing, as newer larger more convoluted models replace the current generation, until and unless there is some additional revolution in machine learning to allow significant simplification.
I recently read that there is an alleged mathematical means to remove linear algebra, and, so, power hungry GPUs, from the process. This possibly reduces costs. . So there is some hope. There's no discussion of power consumption but this seems easily inferred. However, as the AI rush grows, I suspect any power savings will be overcome by adoption growth. Yet almost any savings here, at scale, is a win for the environment if AI adoption were to continue regardless.
Dropping in a suggestion for Zen Browser, a Firefox fork with a bunch of UI customizations that can make it "feel" like Arc - tabs on the side, workspaces with fixed/pinned tabs, side by side view, etc. Comes with more anti-fingerprinting than vanilla FF, but less than Librewolf.
I’m skeptical that whatever they’re building will add more value than Arc’s design does/did. I’ve played with their mobile LLM browser thing and I guess it’s ok, but it feels like a fancy coat of paint on top of the stuff that every search engine is now doing with LLM integration.
I don’t really need an “agent” browsing the web for me or trying to wire together disparate sites (as hinted in the tweet), I need smart ways of juggling tabs, a fresh take on bookmarks that makes those suck less, reduced resource usage, ways to reign in badly behaving sites… stuff like that. Some of this can be done in Firefox userchrome, but the limits of that approach are quickly reached so what’s really needed are new browser projects. Forks of Firefox/Chrome might work too, but the burden of keeping these forks up to date with mainline is massive so I don’t know how practical that is for a small FOSS project in the long run.