I agree that the human brain doesn't task switch effectively, but that is exactly why I want to have many tabs open. In the case of reading Hacker News, I will go through the main page once, opening each article and comment page in a new tab, then will use that as a to-read list. If I am researching how to do a task, I will open have several pages open for documentation, stack overflow posts, bug reports, etc. These are open not for the sake of multitasking, but in order to reduce the amount of working memory I need. For example, if a post mentions that a particular workaround may be required, then I want to keep that tab open until I have verified that that workaround is not needed in my case.
> In the case of reading Hacker News...
This isn't a "task" - you're reading a bunch of completely separate articles, one at a time. You're consuming, not producing/processing/working on a task/problem - and each element of consumption is separate than all of the others.
> If I am researching how to do a task...
I understand and agree with this definition and workflow! This is how tabs were designed to be used. However, I haven't ever seen a common task (for programmers - who have even more demanding requirements than the general populace) that requires a "large" number of tabs. I guess that's kind of vague and I can always change the number, but really, think about it - how often does a task require more than a dozen tabs? More than two dozen tabs?
Just for fun, I opened every frontpage article (as of the time I accessed it - whoops, should have used Wayback HN or something) in Chrome. All 30 articles took 3.9 GB RAM, and adding all 30 pages of comments consumed another 300 MB, so 4.2 GB. total.
> (1) you should only open tabs for the task you're currently working on
In good part because there are no other simple "read later" alternatives in a browser.
Something that would be an option would be a special single use kind of bookmark that gets deleted once you reopen it.
I am on the extreme side of keeping too many tabs open, part of the reason is that they offer a very simple and unique functionality in the browser (it was my reason between other to switch to Firefox, it could easily handles a few hundred tabs (with some restarts from about:profiles))
from what i observe in others, it seems that tabs get created when a new task needs to be done in the same site (that is already open) but dont want to loose context of the previous ones...
from my own personal experience, i like to reuse them as much as i can, but it gets harder as more and more get added, its easier to start from scratch and get a new tab open and go where i want to go...
i think browsers could be doing much more than just tabs and bookmarks to help people organize, because right now it seems like the equivalent of a desk with lots of papers strewn around where someone grabs yet another blank peice to start writing on...
Absolutely. Zettelkasten is the hotness right now. I'd love to see something like a cross between Zettelkasten and Zotero.
Personally I want something in between bookmarks (which I actually infrequently revisit) and history, which is everything and too much noise, and also not permanent in case of history wipe. Strangely I bookmark a lot but don't revisit often, but when I _don't_ bookmark something and need to find some memory thread, it's awful.
I've been scheming about something which I could hit a hotkey, grab the text and metadata, index it, and save the index, set of words/ngrams, or maybe just zipped html content, it's light enough. Then some personal search interface.
Could you give a few examples of this? I still can't think of anything particularly common that would reasonably justify more than a dozen tabs at once...
> i think browsers could be doing much more than just tabs and bookmarks to help people organize, because right now it seems like the equivalent of a desk with lots of papers strewn around where someone grabs yet another blank peice to start writing on...
Yes, I completely agree. I'm working on a piece of software that goes several levels beyond what browsers currently offer for information organization, and (per the topic) it optimizes for CPU consumption at the expense of memory, for the reasons I gave in my comment. However, for the browsers that we have now, bookmarks are much better for organization than tabs. You can tag them, name them, and put them into folders - none of which you can do with tabs.
For example, currently I'm researching bicycles for my mother. I'm on an online shop, and I just open potential candidates on new tabs, which means they start loading in the background while I keep browsing. And using Tree-Style Tabs, this makes them all "children" of the listing tab, which would be lost information if I bookmarked them instead. When I switch to another task, I can just collapse this tree, and when I get back, the listing will be exactly where I left it, which is particularly useful if the site uses JS search without updating the URL. Bookmarks lose all the memory except for the URL and page title.
Bookmarks could be useful for a long-term reference storage (months/years instead of days/weeks), but then chances are a bunch of those links will have rotted, so I need to use something that mirrors the content instead (I used Pinboard previously, now I just use the SingleFile addon to store it as a file).
Bookmarks can be categorized with folders and tags - which, together, are much more flexible than tab groups. A tab can only exist in one group/tree at a time, but a bookmark can have infinitely many tags and exist in a tree-like structure (which is what you described) using bookmarks and be relabeled so it's easier for you to find using full-text search.
> Bookmarks lose all the memory except for the URL and page title.
Yes, because bookmarks are meant to be long-term storage, tabs are meant to be working memory, and transient web page state like whether a radio button is checked or not is not intended to be long-term. Web page state is supposed to either be stored server-side or client-side using cookies or HTML5 local storage - which persist after the tab is closed.
> if the site uses JS search without updating the URL
Then you can just keep that search tab open if you absolutely have to interrupt yourself and go to another task. Bookmarks will do just fine for all the stuff you found. It's extremely unlikely that you're going to have more than a few of these tasks that all absolutely require page state that isn't present in the URL or through local storage and you're accumulating them faster than you can complete them.
> Web page state is supposed to either be stored server-side or client-side using cookies or HTML5 local storage - which persist after the tab is closed.
Supposed doesn't matter to me, frankly. A solution for me must work for most sites, regardless of whether the web developers of those sites do the "supposed" or not.
But even if they did, those cookies and local storage aren't tied to the bookmarks, so I can't rely on them. Chances are they'll expire much sooner, plus they generally only work on the same machine where they were created, which in the long-term may not be the same I'll be using later.
And the ultimate state loser still applies: URLs rot, and bookmarks rot with them. For long-term storage, they're absolutely not enough. Hence my use of page archiving as files, instead.
> It's extremely unlikely that you're going to have more than a few of these tasks that all absolutely require page state that isn't present in the URL or through local storage and you're accumulating them faster than you can complete them.
Maybe, but I don't care to waste time finding out which pages do or do not have their state in the URL or local storage if I need it. Tabs simply work.
I clicked an internal link in this thread (in firefox) and after it loaded (1s) I clicked the back button. 0.7 seconds later I was back but not at the same position. So I had to scroll to find the position to continue to read. I can save 0.7s load time + 5s or 10s of search by opening a new tab instead.
Disadvantages of single tab usage (off the top of my head):
- additional load time
- formular inputs should be preserved, but I wouldn't count on it.
- highlighted text is not highlighted any more
- server can send me another version (or nonthing 404). So I lose the content alltogether.
-
as an example, someone asks "hey can you update the appstore screenshots" and while i may still have appstore open on anothet tab, ill open a mew one just for that task, then later somone asks, hey whats up with jira ticket 1234, now i already have 3 tickets open in jira, and i dont wanna disturb those, somill open another tab, same with online docs like confluence, or web searches on google... at some poont the overwhelming number of tabs themselves get to be too much, so clearing context to yet-another-tab, is a way of coping i think...
i see this all the time in group meetings btw
> I'm working on a piece of software that goes several levels beyond what browsers currently offer for information organization, and (per the topic)
that sounds cool. one idea ive had for a while is that sites (like confluence ot jira, or gdocs) are basically apps, so they should be treated as such and automatically grouped together (im windows or tabs) just like native apps...
By "browsers we have now" you clearly mean "Chrome, the only browser I care about". Firefox can handle literally thousands of tabs with lower memory than Chrome using one or two dozen due to different trade-offs. Very few people ever use that many, but the presumption that no better should be done on the memory front because your favorite doesn't manage it isn't supported by the facts.
Bookmarks aren't used in the same way as tabs by most people and it's telling that your workflow so easily conflates them that you're making criticisms based the rest of us using them differently. I get the impression that what you want bookmarks for would be better done with tab groups for most people? There are certainly a decent number of happy OneTab users out there who're probably closer to your preferred interface than the rest of us, but even on Firefox its userbase is minuscule.
False. I use Firefox for all of my browsing, hate Chrome, and only use Chrome as a whole-browser sandbox for Google products.
> Firefox can handle literally thousands of tabs with lower memory than Chrome using one or two dozen due to different trade-offs.
Yes, and I'm arguing that that's a tradeoff and that's a good thing based on the intent of tabs.
> the presumption that no better should be done on the memory front because your favorite doesn't manage it
Neither did I say that no better should be done, nor do I actually believe that. Presumptuous today, aren't we.
> isn't supported by the facts.
The facts are that tabs are intended to be a working set and are optimized for that. If you don't use them as such, you're going against the design of the tool and expecting it to do well, which is insane. It's equivalent to using files on a filesystem like RAM (expecting extremely quick access and no wear over time) and then complaining when they're slow and your SSD wears out in a few weeks.
> Bookmarks aren't used in the same way as tabs by most people and it's telling that your workflow so easily conflates them that you're making criticisms based the rest of us using them differently.
Correct, because they aren't different. Bookmarks are intended to be used for saving and organizing things for the long term. Tabs are intended to be used for actively working on a task. They're two different things. Thinking that the case where you reasonably want to work on a task that requires dozens of tabs to be open, all at once, all on the same topic, and you switch between all of them very frequently is a common case is completely unsupported by evidence, and logically incoherent.
This has nothing to do with my workflow, and everything to do with the design intent of the tool, and proper use of the tool. Again, see the example about using disk as working memory.
> I get the impression that what you want bookmarks for would be better done with tab groups for most people?
Again, this is something that bookmarks are much, much better optimized for than tabs. Bookmarks can exist in folders - which are probably like tab groups, but first-class, and much better supported - and tabs cannot. Bookmarks can be tagged, and tabs cannot. Bookmarks can be exported to a file, and tabs cannot. Bookmarks can be renamed, and tabs cannot. Bookmarks are straight-up better at their job then tabs are. This is not "better done with tab groups for most people" - this is "worse than tab groups for everyone".
To summarize: working memory is not the same as long-term storage (in the human brain, in operating systems, in physical desks, in the browser, and everywhere). Tabs are for working memory, and optimized for that. Bookmarks are for long-term storage, and optimized for that. Bookmarks are objectively better than tabs for long-term storage in non-performance metrics (can be organized using folders and tags, can be exported/imported, can be renamed) and performance metrics, and you and everyone else who insist on using tabs like bookmarks are (1) doing yourselves a disservice by using the tool wrong and suboptimally and (2) doing everyone else a disservice by demanding that the tool literally be made less optimal to fit your incorrect use of it.
How users use the program determines how the program should be implemented, not the other way around.
My anecdotal observations are that this often causes extreme number of tabs. I've seen multiple examples from social media in the last month of people scrolling through their mom's (/little brother's/etc) phone, showing hundreds of tabs, but usually all duplicates of the same dozen or so. I remember seeing similar things firsthand (but not to as great an extent) when the first generation iPad came out. And I've also witnessed the same in offices, where doing something in an app or an existing page will arbitrarily open a new window or tab, leading to an explosion of windows and tabs all hidden beneath the IE11 icon in the Windows taskbar.
By all signs, it seems true that (a) average people don't know what tabs are, and (b) they have a lot of them. So it makes me wonder about your methodology. I used to be a Mozillian, and I remember the less-than-rigorous excuses that Firefox UX folks reached for when backfilling justifications for their design whims.
Honestly this is the only reasonable way to use tabs on a mobile browser. There is almost no drawback of keeping them and closing them is a chore. I get it sounds like the typical "boomers using the internet" but given that you should expect old inactive tabs to unload anyway having even hundreds of tabs open is a detail.
That said, what you're saying is, "In practice, users don't have a lot of tabs open", which is very different from what I was responding to, which was "users shouldn't have a lot of tabs open".
Of course important to remember that increased memory utilization results in increased CPU utilization. The range of degradation is quite wide: Increased cache misses (which cost a significant number of cycles), access patterns not in line with the prefetcher, and increased TLB misses.
It's for good reason that one of the most potent optimization techniques for hot code is to reduce the memory footprint and memory access.
Rather the trade off you're thinking of is "caching vs. realtime computation" rather than just "memory vs. cpu". Outside caching, it's pretty universal that more memory utilization just means poorer performance, both in wallclock time and in CPU utilization.
It's also not very useful when someone else is hogging all of it! There is an opportunity cost to the user and other programs. It is supremely selfish to only think of oneself, and reserve all the common resources because "well it was just laying around".
I PAY for the capacity in case I NEED it, not for some bloated browser to snorf it all up because they can't be assed to write reasonable software.
A common use-case: I visit a site I open a new link in a separate tab. Those tabs are visited later.
Having gmail open, slack web client, jira, git and the dev and prod versions of a site I'm building is a common scenerio.
Use the great suspender extension; my friend in academia thanks me every single time we meet. And now I see them have 4 chrome windows open with each having ~20+ tabs in it!!!
For them everything happens in browser, its synced and the great suspender lets them run that monstrosity without choking their MacBook Airs
One problem is that CPUs have various ways to use less power or even switch off parts when not in use, but you always pay the same for RAM.
For battery-operated devices that sleep most of the time such as smartphones, that means being careful with memory usage is the way to go. I think that’s part of the reason iPhones ship with less RAM then Android devices (reference counting needs less RAM than garbage collection for the same performance level)
Additionally, the extra power cost of adding more RAM to a device is very low relative to even using 25% of total CPU TDP. [1] gives 3W per DIMM for DDR3 (which is an upper bound, given that LPDDR3 will consume far less power while fully on, let alone in deep sleep mode) and [2] gives 2.5W for DDR2.
Edit: aha, found a paper[3] on RAM power consumption in a phone. Far less than 5% in either "suspended" or "idle" states, under 10% in almost every other benchmark, and universally lower than CPU power consumption.
[1] http://www.buildcomputers.net/power-consumption-of-pc-compon...
[2] https://superuser.com/questions/40113/does-installing-larger...
[3] https://www.usenix.org/legacy/event/atc10/tech/full_papers/C...
This is generally a good trade. Memory is not very useful (modulo filesystem caches) when you're not using it, and has a very low cost when you are (extremely tiny power consumption). CPU cycles cost power (leading to heat dissipation and battery life loss) when you use any amount, and latency when you use too much. Latency directly affects user experience and efficiency, which is a much higher cost than "my RAM is at 75% usage".
Additionally, Chrome's tradeoff is such that it works out better on cheaper devices. Memory is far cheaper than more powerful processors - which also reduce battery life, which is even more expensive. On more expensive devices, you have all the memory you need for Chrome anyway, so you're just getting better latency.
Finally, there are very few cases where you should have a large number of Chrome tabs open anyway. Multi-tasking is impossible (the human brain simply doesn't support it) and rapid context-switches are inefficient. As much fun as it might be to have 5 tabs open on each of 10 different topics, you're simply decreasing your productivity and capacity for focus. Even when doing multi-disciplinary research, there are a ton of ways to get around the problem of running out of memory due to Chrome tabs: printing pages to PDF, turning off JavaScript, downloading research papers and reading them in a native PDF viewer, extracting content from webpages and inserting into local documents, and the classic using darn bookmarks instead of tabs like they were intended to be used.
(of course, in an ideal world, all web browsers would use absolutely no memory or processing power whatsoever - I'm just defending this particular engineering tradeoff)