Some news from LWN
lwn.net
lwn.net
I was able to find the link from the front page, but it's kind of hidden -- https://lwn.net/subscribe/Info which says "Get started! The first step is to log into LWN. If you do not yet have an account, please create one now; otherwise, please log in to continue:". Which seems like a lot of steps to pay them money.
I ran into this gem recently:
What an odd name. Flog rappeling? Flo grappling? Eff log rap pling?
/me loads page
Ah, door #2.
There's nothing wrong with informing people that you can subscribe! A small Call To Action at the end of every post (Do you like this post? Help support us and get future content directly to your inbox by subscribing) would already do wonders! And having "Subscribe" in the sidebar. (If anyone from LWN is reading: use those words verbatim if you wish)
There's no shame in this being a paid publication. Considering the anti-advertising demographic, many would be glad to help support an ad-free system.
Okay found it in the middle of a paragraph on the front page. I think the nav could be improved.
Not saying they should introduce annoying "subscribe now!" modals, but they definitely could sell it a little harder.
For example, https://lwn.net/SubscriberLink/695910/61948e743f7054ec/
It is paywalled for a short period (2 weeks??). After that, everyone can read their articles for free.
> LWN is, of course, a [subscription]-supported operation.
But I agree that's it's not a strong call for action, and that it's hard to search for if you overlooked that sentence.
I assume this is some sort of principled stand designed to avoid skimming money from people who no longer use their subscriptions but are too lazy to cancel... but it probably means LWN is failing to accept money from a lot of people who would be very happy to be paying.
Don't be afraid of hyperlinking; it's the reason the Web was created, after all.
Apart from kernel development, it did excellent technical reporting on systemd, prior to it being widely adopted and it does a great job of keeping you in the know of how things like wayland are actually progressing.
As for the parts that are genuinely over my head, I've been surprised how many times I've read about something on LWN and not really understanding what I was reading about at the time, but connecting the dots later.
You know when you read something and have no use for it, so you delete it from your front-end memory, but it resurfaces just when you actually need it?
Just to give an example, I remember reading a series on LWN some time ago about how virtual memory is managed in the kernel and now that I am writing a toy OS in Rust, I regularly have flashbacks to that series and it makes the concepts that I am learning about now genuinely easier to grasp, because I can make an instant, be it clouded connection that I did in fact hear about this and this already, even if it didn't make a lot of sense at the time.
There's also been times when the kernel has been misbehaving on me and I remembered that there was an article on LWN talking about this very issue and how adding a certain kernel boot parameter may resolve it, and it usually does.
For me, LWN is well worth the $7 a month they ask for it just on these points alone, it is one of a very few internet publications that in fact I am willing to pay for and consider it money well spent.
"LWN: A site you want to read the comments on."
I think the main reason of the drop in readership is firstly kernel development is very involved and detailed, LWN mostly reflects the kernel mailing list and associated events and tech talks.
I guess what I find most difficult as an outsider of the kernel community, is trying to keep up. Jonathan Corbet is doing and extraordinary job summarizing the activities of the kernel community in LWN. But the kernel doesn't maintain any main "Epics" that we have in software traditional development, so take for example fair scheduling. AFAIK, the kernel community does not decide to tackle fair scheduling on the next release of kernel, ie 4.7. Then say power management in 4.8 etc...
LWN ends up with a smattering of various articles under common themes like "Kernel", "Security" etc...
LWN has in the past explained long running topics like scheduling when key updates are happening and I commend them for that.
None of the above is a negative reflection on the kernel community or LWN in how they operate but it feels kinda hard to keep up and stay vested.
Anyways just a few thoughts, will renew membership now.
If the kernel has anything like an epic, there is multiple concurrent epics because the Linux is not a monolithic project. Nor are epics related to any kernel version.
For example, each major GPU driver (radeon, novaeu, intel) manages their own development, and push things upstream. DRM (the kernel part of DRI) interacts with those guys, and DRM pushes things upstream.
Network drivers and the network core are managed by small teams, push things to each other (such as adding support for major intrusive features like, say, when they added segment offload support, or recently genericfied the interface for that to support more protocols; or alternatively, when they started making more complex DMA patterns like scatter-gather and other zero-copy methodologies happen). The network guys push things to each other, and push things upstream.
The storage guys also do that, storage drivers and storage core push things to each other, and push things upstream.
Every possible major section of the kernel is a project in of itself, and have project leaders, and have parts that are ran by different people.
Generally, you don't push a patch to lkml by itself, and then you're done; this almost never happens. What happens more often is you go to one of the people who manage that part of the kernel, or has significant experience with that part of the kernel, and discuss your patch with them, and then they will try to help you make a patch that can be merged upstream.
More patches are rejected than committed because of how this works. Linus, or any other kernel maintainer, does not want your patch if it hasn't been ok'ed by the people who maintain those other parts of the kernel unless it is a small patch that does something obviously useful.
And what about the companies that just push drivers? Companies employ kernel hackers that know how to write drivers. For example, Intel employs the Intel GPU driver guys, Radeon employs the Radeon GPU guys, and all those guys are already major authors in the X/DRI/Gallium/DRM/whatever community; the infrastructure to make that all work is shaped by the hardware, the hardware is not shaped by the infrastructure.
The thing is, the Glory Days when kernel development was a scary, exciting wild west are lone gone.
By necessity, the Linux kernel is now a part of the tech firmament and development is necessarily not as sexy as it used to be. So it's gonna be tough to get people to pay to read about it.
It just mirrors internet culture as a whole. Easy to consume information, show off and fashion.
Lest aspiring systems programmers get the wrong idea: systems knowledge is still black magic and it's still plenty valuable. Knowing how systems actually work under the hood is a rare skill and can help you code circles around the node-is-close-to-the-metal [1] competition even when writing application code.
[1] https://twitter.com/shit_hn_says/status/234856345579446272
For example, as of right now, most of the articles on the front page aren't about kernel development: distro security updates, LibreOffice, Firefox, GnuPG, ...
I think this is perhaps an overlooked aspect of Linux's success; at a critical time, they had LWN and good documentation that made it easy for new developers to understand what's going on and start contributing. I don't know any other open source project that is as accessible to new developers as the kernel was in the early 2000's.
I think in the last few years many open source communities have become a lot better than the kernel in terms of having good starting points for beginners. In the last 5 years I think many projects have started to focus on developing and improving community process, which is good.
I must say that this website comes across as very unprofessional due to the total lack of any design elements which would make it more easy to consume the content (spacing, choice of font). Also it seems like they don't use CSS to enhance the site's UX at all. If you look at the mobile site, they have some sort of dropdown to show the menu, which is then also way too small to properly click on. Issues like this are allover the site: Tables without any proper spacing / borders, mixing centered and left-aligned content with no reason, etc.
For me it just feels clunky and even though the information they aggregate seems to have some value for others, the presentation is just way off.
Underneath each comment's title is a full-height line with a font stating the UTC timestamp, and in the next line we can finally read the actual comment.
I can't be the first person to criticise those issues, and from the fact that they haven't changed it already I think they actually don't care about these kind of things.
I didn't downvote your post... but I relate to the frustration that when we DO get content in the former category, the top comment is always a critique of website layout or font kerning or something else that tries to pull it into the second category.
Can you just... not?
It most certainly is.
You acknowledge in your parent comment that you are not the target audience for this site. I assure you that the sort of crowd that's into Linux kernel development by and large prefers web design that other audiences would find archaic.
Conversely, were LWN to have switched to Bootstrap five years ago, or some Medium-like look today, or whatever the next transient fad will be five years from now... their audience would by and large viscerally hate it.
Designer types indulge too much in the fantasy of OBJECTIVE standards for "organized", "hard to access", etc. It's all subjective. Hence, fashion. For people who enjoy working on green-screen terminal applications, switching to a web app version is like giving their workflow a lobotomy. For people who prefer mobile-first minimalist web apps, the reverse is true.
LWN, and it's related subculture, happens to not be your scene. That's okay. But it's perfectly fine to just, well, not comment. Because it is so painfully lame how whenever a deep technical topic comes up on HN, half of the comment section deals with website aesthetics. Hey guys, if some topics just aren't for you, then it's... just... not... for... you... and... that's... okay.
Really? I've always viewed plain HTML sites in the complete opposite way. They tend to have really good, well thought-out content. There's no filler or clickbait with them.
On LWN there is unnecessary and redundant metadata accentuated!
Everything on HN is clearly thought out, and almost nothing on LWN is.
"N minutes ago" is compact and does not require any client-side JS to provide an unambiguous value.
At least LWN includes the timezone, which is a step ahead of most of the sites I've used.
Ditto. Maybe I loaded the page yesterday. What does '1 hour ago' mean? Was it updated by JavaScript? Is it an hour from now, or from yesterday, or what?
Give me a timestamp, and I know.
I love that phrase for all it implies about the point of view of the user thereof — to my knowledge, every site is an HTML site.
Excessive CSS is, well, excessive. That said, a tiny pinch of CSS goes a long way towards making a page readable and aesthetic.
For me that's a feature, not a bug. I'm interested about the content itself, not about with what color or what font that content is presented as.
The article's title is repeated throughout the comments list and emphasized. The other comment metadata has the same visual emphasis as the actual comments (except the link and author jump out more than the comment itself). The articles themselves expand to the full width of the window.
There is plenty of introductory material on visual emphasis around, not just covering emphasis in textual media, e.g.: http://daphne.palomar.edu/design/emphasis.html
I agree that line length according to more recent studies seems to be less consistent (some recent studies say speed actually increases with longer lines but there doesn't seem to be any data on comprehension/retention and preference seems to be biased towards shorter lines -- again all of these factors may be cultural). But visual emphasis is a thing -- it's hard-wired into how human visual perception works.
Most techniques for making metadata less emphasized make the metadata objectively less readable. I like having the metadata there. I like being able to read it. I have no trouble reading the text (because it's a decent size and the layout handles zooming without issue.)
Also, the title of the article is repeated throughout and drastically emphasised (considering nothing else is that visually distinct and it even has a solid background color).
Using a line break between meta data and content is also a familiar paradigm. Spend 20 years reading your email with mutt or mailx and you'll catch on. ;)
I don't know what the original commenter meant, but the text could be a bit bigger and the line lengths and spacing controlled.
The comment metadata could be shortened and humanized, though the audience may find it useful.
There's some bugs with the dropdowns on mobile (haven't been able to close one).
I don't know if that will get them subscriptions, but it'd make the site easier to read.
Ironically, if you look at the HTML classes, they are using a CSS library typically used to make "modern" sites many here revile: http://purecss.io/
Good design makes the content more evident and less tiring to read. The two go hand-in-hand.
Do you favor poorly written articles simply because you're interested in the "content"? I expect not, so why the latent notion that "function = good" and "form = useless"? It's intellectually lazy.
Say what? On the contrary, I think that it's splendidly simple to read LWN, because there's no distraction. It's a good, clean site; it respects the styling I instruct my browser to apply to it; it displays just fine in links, lynx, elinks, ewww, emacs-w3m, Firefox, Chrome, Safari, Internet Explorer, Edge — if that's not professional, then what is?
The HTML is very clean and readable. They do use CSS.
Granted, it doesn't look trendy. Indeed, it looks like something from the 90s/00s. But once one updates to a20`6 style, one then has to spend money in 2017, 2018, spring 2019, summer 2019, fall 2019, winter 2019, January 2020, February 2020 — one will be paying designers and artists until the end of time.
Would I appreciate a somewhat more current look? Sure. Would I trade good HTML and universal readability for a more current look? Hell no.
2016 was a total mistake. Let's just revert, and then wait for 2020-LTS.
Argh, I blame an unfamiliar keyboard. And being half-asleep. Sigh
Granted, the current trend to make everything feel like a shitty JavaScript application would be taking it too far, but lwn.net is definitely not a pleasure to read. Yes, it displays "just fine" in links, lynx, elinks, etc., if your definition of "just fine" is Fefe's Blog.
Hey, fefe has some pretty nice stylesheets for his blog
https://blog.fefe.de/?css=darkgreen.css
https://blog.fefe.de/?css=cold_blue.css
https://blog.fefe.de/?css=ascii.css
https://blog.fefe.de/?css=topaz.css
https://blog.fefe.de/?css=janvsfefe.css
This is how it looks if you visit the site without a CSS cookie.
It's clean, very simple and renders fine in everything, even mobile. No distracting content, proper font and font spacing.
This would probably fail a basic Universal Accessibility test based on that alone.
> Nathan Willis, who has been an LWN contributor for many years and an employee since 2012, will be stepping down at the end of September to pursue an unmissable opportunity to study one of his non-journalistic passions: fonts and type design.
Related?
Nate is not going to be working on the fluffy parts of typography.
And you're commenting this on Hacker News. A site built up on tables, inline style-definitions, and where it for half a decade refused to add a simple...
<meta name="viewport" content="width=device-width, initial-scale=1.0">
I'm not saying it's not a factor, but clearly this seems to be an issue you're able to overcome, given the right circumstances and content.
Also, I'd like to think of LWNs look and feel as a "solid" and clean web-page, the way they used to look and work back in the days when HTML was simple. Nothing excessive, and no overdone CSS/JS madness, which increasingly seems to be the norm.
You just point your browser at the page, the browser retrieves the HTML, the HTML contains the content, and the overall wastefulness of network bandwidth is way below 10%. It's the way I wished the web still worked.
I have never thought much about the design of the site and that's a good thing. It succeeds in delivering information quickly and efficiently, which I imagine is the most important thing for LVN's readers. I would consider almost any new UX feature a regression if there is even a small chance it breaks my browser or slows down page loads.
You're right the LWN site is utter trash. It's 2016, why aren't they embarrassed of their own incompetence? What comes next? Are they going to tell me to read a book?
For me, "design elements" tend to make content harder to consume.
> it just feels clunky
Pages that load megabytes of JavaScript to display kilobytes of text feel clunky. The modern trend toward large icons, large fonts, wide margins, and slow response times because everything has to go through JS/XHR makes websites that keep with today's design fads feel like toys, while websites that present information without a ton of JS and CSS baggage feel like tools.
I feel like design fads are partly a clever marketing ploy to keep web designers in business -- "it doesn't look modern" shouldn't be a justification to plop down the big bucks to redesign a perfectly functioning website every 3-5 years, but it is.
> large icons, large fonts, wide margins
I have 20/20 vision but even for me, all these are boons. It's the "design types" that want small fonts because they look cool. But for reading, give me a 14 or 16 font any day of the week, much more relaxing on the eyes.
All those things you're complaining about improve readability and discoverability.
I can't even imagine how people with bad eyesight think about this topic...
> and slow response times because everything has to go through JS/XHR
This does, indeed, look like a (software) design fad.
I wouldn't say that it comes across as unprofessional, though — rather the opposite. To me, LWN appears to be professional: no-frills, focused on substance over form.
But the kicker is, whether or not someone thinks the site looks professional, it's a subscriber-based service that's been running for nearly two decades, that has hardcore professionals subscribing to it. It's not FOTM.
LWN gets me what I want. It is the most important resource for a linux kernel or in general Free Software enthusiast. Average HN crowd perhaps does not fit into that stereotype.
If you perhaps not enjoy LWN enough to subscribe - at least consider if you have any friends that might be interested or persuade them to subscribe - if they're already readers.
Well researched information is an important way to avoid incorrect assumptions to spread. Thereby hopefully avoiding bad decisions to be taken, as well as actions on problematic things (e.g. gmane).
- Improve the design a bit. Carefully, you do not want to piss of old subscribers.
- Improve the conversion rate. There is a middle ground between newsletter modals and burying the method to subscribe.
- Introduce a few new products. As far as I can see, at the moment it is just "in depth news for OSS, especially kernel hackers". Maybe they could package some parts of this up for different audiences (security notices, content partnerships etc.).
- Some marketing tactics. "Invite a new subscriber to get this badge." SEO could be improved, too.
I simply tried to change the font-family to Segoe UI (or whatever your OS' font) in Developer Tools, and it looks more interesting than Times New Roman
Essentially increasing the visitor time for new visitors, then increasing the recurring visitors, then increasing the sign ups.
Gamification, i.e. badges, I don't think would go over too well with their subscriber demographic. I think something like a free month, a free public share of a link (I don't know LWN very well, but I do remember reading articles there which say that some member sponsored the article to make it public) or a bundle of $RELEVANT_TO_YOUR_INTERESTS white papers may work better.
Otherwise I think you make some very solid points, especially trying to find out how to subscribe was much harder than it should be...
Which would then have to be displayed, which would be more visual spam for no good reason, which would piss off the core users.
I cannot read the kernel source code as easily as the great articles on LWN that lovingly explicate the details. Sometimes even the articles are challenging for me but I learn something new every time I visit it.
If HN and the now nearly defunct Slashdot are popcorn, LWN is fine organically grown fruit and vegetables.
I am off to subscribe to LWN and urge you to do so if you find the kernel interesting
They also cover lots of non kernel topics. There was a fine article on the software modifications usef by VW to cheat the pollution testing, for instance.
Either way, I've gone ahead and paid for the year up front, please keep going, I thoroughly enjoy the articles!
that is being enabled by your subscription!
* 6 x LWN.net Weekly Edition for $DATE
* A few weekly security advisories
* "Some news from LWN"
* Statistics from the 4.7 development cycle
* New software releases
None of these interest me. Here's what does:
* Klitzke: Why Uber Engineering Switched from Postgres to MySQL
* Python's os.urandom() in the absence of entropy
* One-time passwords and GnuPG with Nitrokey
There are a few articles I like there. But they aren't discoverable and I can't find what I want without working to filter the noise.
Contrast this with hackaday.com, where every article is a really interesting feature, there's no news on bugfix releases or kernel changelogs, and they mix feature news, cool stuff in the store, and community projects.
lwn could do better though, make the website visually better, have a mobile app, do some beginner friendly kernel corner and such, also add a save-my-favorite-articles function under my account(last time I checked there is no such thing), adding tag support,etc. In fact just moving to wordpress will probably solve many of its website "issues" so it can focus on the content as it has been for a long time. You need the landing page stay up to date to attract new and young subscribers.
Notice that it's seven printed pages full of questions that somebody familiar with the thing might want to know. But nowhere does it tell you what "LWN" stands for.
If you want people to subscribe to your thing, step one should be to tell us what it is.
"What does LWN stand for, anyway? LWN, initially, was "Linux Weekly News." That name has been deemphasized over time as we have moved beyond just the weekly coverage, and as we have looked at the free software community as a whole. We have yet to come up with a better meaning for LWN, however."
I'm happy to keep supporting them.
(Since others have commented on the design/redesign I'll say one thing: The way the sidebar works now means I can't zoom text in on my iPad because it gets covered by the sidebar. It's a serious functionality dent which I wish would be fixed. That's it.)
Keep up the great work.
really? The font size is smaller here than on LWN, and the width of the text is smaller on LWN than on HN.
Comment metadata is slightly better on HN, but largely because on LWN you can put a comment title and here you can't.
The poor text contrast in many places makes it borderline unreadable, text posts in particular.
My second complaint would be that line lengths are often too long to be easily readable.
1) The front page for logged in users changed to a less useful one-column layout, so now I always browse the site not logged in.
2) Previously the current and previous weekly editions were nicely visible in the sidebar. That isn't there any more. As a result I read much less of the actual articles, as they just don't catch my eye. I mostly just check the front page news column now.
Anyway, I hope all the best for LWN, as the content is of very high quality indeed.
At a couple of places I have worked at we had a group subscription for the department, $45 it not terribly hard to get approved. I think its a great perk to list for prospective candidates too.
Maybe they don't want to refocus away from the public, but it could work.
I'm proud to have been a subscriber for the past 4 or 5 years.