This is kinda like a mini-blog where I gradually present the new approaches in the 2022 version of writeups.org (WORG).
Because doing it all at once would be insufferable.
- Episode zero – Men at work.
- Episode one – The basics.
- Episode two – The Cookies and ads.
- Episode three – Fear will keep them in line. Fear of the commission.
- Episode four – Dancing in December.
- Episode five – Time to read, more or less.
- Episode six – Voice of the worg.
- Episode seven – Meticulous.
- Episode eight – Trifle.
- Episode nine – The glossary strikes back.
- Episode ten – I’ve got the Powers.
- Episode 11 – Return of the roving ranger.
- Episode 12 – For those about to print.
- Episode 13 – The homepage shuffle.
- Episode 14 – Let loose the poodles of strife.
- Episode 15 – HREF hygiene.
- Episode 16 – Better images (1/3).
- Episode 17 – Toward the ides of September.
- Episode 18 – Seeking search.
- Episode 19 – Better images (2/3).
- Episode 20 – Categories in the craddle.
- Episode 21 – Everything counts.
- Episode 22 – Top Ten.
- Episode 23 – Better images (3/3).
- Episode 24 – Micro-moolah.
Episode zero – Men at work
Preparation work is still ongoing, as the site will have another major evolution in layouts (plus some functionalities) in 2022.
As usual, we’re working with a very… modest budget. So there are side effects, little glitches, a temporarily pared-down search engine, the pages charge a bit slower than they should since the caching is temporarily turned off, etc..
There’ll be more details when the changes get more concrete, of course. But the general idea is that the bulk of the evolution will be another slow-burn, entry-by-entry process that will take years to sweep though the entire site.
This work will likely leave us slightly in the red, financially. Nothing terrible, but still – in the red. So if you’re among those who still have a disposable income, you can consider donating.
That’s all for now, more on the next episode.
Episode one – The basics
A new version of the site is now being deployed. Let’s call it WORG 2022, though WORG 2022+ would be better.
“2022+” would be better because I expect the full rollout — that is, every single article is upgraded to the new format — to take about five years. Maybe more.
The big ideas of this update are :
- Finishing stuff we lacked the money and time to do with WORG 2016.
- That includes better displaying on smartphone. What we had was okay-ish, but it’s been years now where most traffic comes from smartphones. So it needed to be better.
- That also includes the search engine, where we had to just… stop rather than wrap things up.
- Having a more colourful, bolder, more structured look while still retaining a minimalist vibe.
- Improving loading times, though they already were fairly good… most of the time.
- Making it easier to look at the images, while solving the age-old problem of “we want to have big, HD images but also to display on so-so smartphones on so-so networks”.
A key point — the articles are getting upgraded to the new format by hand.
Said upgrades will go from the most recent article to the oldest. So at first it won’t be thrilling. Since I’ll be working on recent articles you probably already read. Changes do take place, but it’s mostly copy-editing.
From there the pace will gradually slow as articles require, to be upgraded, more work. Better images, more corrections, updates, edits/rewrites for concision and clarity, etc..
Which is why it’ll take years.
And so yes, during those years the site will have a mixture of new-format and old-format articles. This might be familiar from previous lengthy, site-wide WORG upgrading runs.
That’s all for now, more on the next episode.
Episode two – Cookies and ads
I’ve updated our FAQ page about the cookies and ads used on the site.
Normally I wouldn’t mention it, but this part of the FAQ saw crazy traffic – like, more than 10K readers. I just don’t get it. Did a large site link toward it as an example of what not to do ? Is this a web-crawling robots thing ?
That’s all for now, more on the next episode.
Episode three – Fear will keep them in line. Fear of the commission.
One of the new things is to start articles with the website equivalent of a splash page. It looks cool, and it’s an efficient way to convey what we’re going to talk about.
There’s also a splash image on the soon-to-be-finished homepage. For now, it’s a cheap placeholder. It says, in a visual manner, “here is what this site talks about”. But it says it in a cheap way.
This will eventually be replaced by a much better image. Do you want to know how much this commission costs ? No. You don’t want to know how much the commission costs. The cost of the commission is horrific. The cost of the commission will break your gluteus maximus.
The cost of the commission is expressed in real-world, grown-up money. Not on the writeups.org-scale, “one shoestring, a small but pretty common rock and a broken elastic band” kind of money. It’s actual, palpable moolah. Wearing a suit.
Raising funds via donations to cover the cost of the commission would be like peeing on a three-alarms fire.
So I’m raising funds in a different way, which I promise doesn’t involve selling illegal drugs. Or, worse, NFTs.
And one day, the nice image will come.
Episode four – Dancing in December
The ongoing WORG 2022+ work means that I have to skip a site update – for health reasons. That’s the update that would have loosely wrapped around the end of the month.
(*Technically* it will take place because I will revise and upgrade FAQ pages. But I don’t think that most folks would consider that it counts).
This will chain with the one update that we always skip – the New Year One.
So normally that’s the point where I’d say something cheerful about December and New Gregorian Year celebrations but heh, heart’s ain’t in it. Try not to die, especially if you’re part of the countries and populations that Important People have decided to just let take the brunt of the pandemic.
Episode five – Time to read, more or less.
One small new feature is indicating a reading time atop the article.
A *rough* reading time.
It’s rough in part because developing a more accurate estimate would quickly cost far more than it’s worth. And in part because it has to be rough :
- About 50% of the readership is in North America, the other half is widely dispersed over the planet. That means widely varying levels of English fluency, and therefore reading speed…
- … plus people who use in-browser translation.
- Reading speed can considerably vary depending on device, ambient lighting conditions, etc.. Yeah it’s a platitude, but it’s meaningful in a majority-smartphone era.
- We have no idea how long people look at images.
- More importantly, we have no idea how long people look at the RPG stats. Most probably don’t look at them at all. Many may just quickly parse them. Some might peruse them in detail and check the rules (say, because they’re learning to DCH or to DCA).
And game stats can make up a good amount of the word count of an article. Bird-Man of the Ani-Men would be a good example.
Plus, it’s well known that most people just sort-of-speed-parse online texts rather than actually read them.
So – *rough* reading time. YMMV.
Episode six – Voice of the worg.
One of the small new features in the migrated-to-the-new-layout articles is a player to read most of the article aloud. It’s discreet for now (though it’s more visible on smartphones) since we don’t know how useful it is.
The tech is good. It’s the usual thing where megacorporations let you use their good stuff for free since they need a lot of data to train their neural networks.
Between the negligible costs and being quick to do, ain’t no reason to try it.
The use case is similar to the way many people listen to podcasts. Whilst exercising, whilst doing the laundry or other chores, during short drives or bus rides, whilst doing boring grind in a video game, etc..
It also turns out that having a voice clearly reading the article aloud is genuinely helpful in copy-editing.
Nothing major, but some people will find it useful. So that’s nice.
Episode seven – Meticulous.
So, there are logistical motivations behind WORG 2022+. Such as our old two-column display not being a smart layout on small screens (like smartphones). Or finally finishing the search engine results page. Or sundry budgetary, ergonomic, design, etc. reasons.
There’s another, darkly and sinisterly secret hidden reason.
I’m *never* happy about what’s on this dang site.
- Some images are bleh.
- There are typos.
- Some stats aren’t well-explained.
- Too many sentences have, like, a missing word. Or a weird grammatical bit (probably from having been rewritten a few times).
- Some technical words aren’t explained.
- Some entries are way too verbose and/or unclear.
- Missing tags.
- My bad health means I occasionally zone out and miss issues with the way the information is presented.
- Some stats still use older versions of original rules we have since improved.
And so on and so forth. Plenty of stuff’s not good enough.
And yes, I’m the kind of person who’ll restart a computer RPG 16 times in a row because my character build isn’t perfect. Or because I forgot a dialogue option in the third house to the left. Or because my character’s hair doesn’t look right in playable-world lighting.
So WORG 2022+ is a good pretext to go over EVERYTHING across 6,100+ articles.
One unexpected ally is a new feature. The (semi-discreet for now) ability to play a voice synthesis version of new-format articles. It’s genuinely helpful in noticing small issues with texts.
Episode eight – Trifle.
Not posting much around Christmas. As I assume most people have better things to do than read this.
I’m creating a number of new posts, which I assume are relayed by the sharebots – the automatons signalling on Facebook, Tumblr, etc. that new posts are up. But these pages are archives. It’s very old content about the site — not profiles — which I’ve been reorganising.
As with every site overhaul, I’m taking the opportunity to clean some minor things I don’t normally have time to address.
Episode nine – The glossary strikes back.
Another small feature is adding tooltips in the articles, explaining vocabulary. Usually, technical jargon such as “brick” or “expy”. But sometimes it’s literary/rare words, regional terms such as Britishisms, etc..
This is useful to nearly everyone, but mostly for :
- Our many readers who speak English as a second language. As an added bonus, this means they get to learn stuff. And we get to use more complicated words from time to time, rather than always go with the simplest term.
- People who aren’t grizzled super-heroes and TTRPG geeks.
This is part of a push toward anti-gatekeeping that we have slowly deployed over years. The “Context” sections, the glossary, beginner-oriented FAQs, hyperlinks toward a recap when we off-handedly mention a comic book crossover event, those small italicised paragraphs with additional explanations, hyperlinking game elements that aren’t in the rulebook, etc..
In practice :
- If the explanation is simple, you just mouse over to display it.
- If the explanation isn’t simple but it’s a term we use repeatedly, you get a link toward our glossary.
- If the explanation isn’t simple and it’s not a term we use regularly, you get a link toward, say, Wikipedia.
Episode ten – I’ve got the Powers.
The articles about DC Heroes RPG Powers were reorganised over the 27th and 28th of December. There were micro-edits, but this was almost entirely about logistics and convenience.
Everything now points toward the “base camp” article about Powers. Whose key feature is direct hyperlinks toward every single Power, so you don’t have to know where stuff is.
Episode 11 – Return of the roving ranger.
I recently 2022-ified the character profile for Gitane, a sample Everquest character. I picked that one because it’s a tough entry to work on, and thus a good “worst case scenario” to estimate work durations.
Here’s what I mean by “tough”, which also gives an example of the work going on :
- Entry was slightly too long, was split in two for slower readers.
- Images for the maps were too old, reacquired them by scanning – then enhancing colours for on-screen reading comfort.
- Most screenshots were too old and were of course low-res. Did upscales, vibrance fixes, then used AI denoisers and re-levelling to create the illusion of some additional resolution.
- Ran my javacripts and regex-based PERL for most markup/shortcodes conversions to the 2022+ format.
- One “generation one” tag was missing, added “generation two” tags.
- A slew of links to external links to fix (our links weren’t done quite right until now) and check. Added a few links whose absence was odd.
- A particularly high density of images, which had to be removed then re-placed between sections.
- Re-processing all “right-hand column” shots so they’d fit into the new single-column format. That involved cropping, a neural de-artefacting pass and some fixes (such as excessive blue reflections on irises in one portrait, or denoiser-induced artefacting on tree bark).
- Replaced all the old versions of the images. Crafted the lightboxed/zoomed-in/big/high-res version of each image
- Metric listings for vitals.
- A few enumerations hadn’t been converted to lists yet.
- The entry-specific video had already been shot, so the first half was good. Had to think about and find the right kind of music for the second half’s soundtrack. If that means somebody develops an interest in Bratsch’s music, then that’s good.
- Did the glossary tooltips and links, including some new ones (not every reader knows what “volubility” means).
- Quoted and credited two extra, vintage screenshots.
- Fixed a few typos. Including discreet ones I only noticed when doing the voice synthesis pass.
- Added details to my aside about real-world European nomad cultures terminology.
- Added some links to technical DCH articles (weapons, armour…) and to another Everquest article (wot explains Barbarians).
- Routine copy editing about long/unclear sentences, long paragraphs, missing words, etc.. This entry was recent enough to have had decent passes already, but my character concept notes were too off-the-cuff.
- Adding the new “video link” glyph to the link toward the (excellent) excerpt of the Carmen opera. It replaces my old, ugly notation for video links.
- Tweaked the DCH and DCA stats as usual (the previous display wasn’t great for smartphones, particularly the DCA attributes).
- Added some tiny details, now that the entry is split and therefore not too long anymore.
- Generated and checked the voice synthesis files, using a voice that felt more appropriate for the character (but turned out to be too robotic-sounding when reading long texts). Added a pronunciation note.
- Added my writer avatar 😺.
So that was a solid afternoon of work, for just one entry. One that had an editing/polishing pass just a few years ago.
Episode 12 – For those about to print.
It was implemented weeks ago, but as a reminder – all entries (2022+ or pre-2022 format) now have revised printing or print-as-pdf buttons.
Them’s small buttons near the top.
Printing has been a niche use for years, but when it’s useful then it’s useful. However, different folks gonna need different things. Say, just the DCH stats or just the DCA stats, some pictures or zero pictures to save ink, etc..
So we just offloaded the printing thing to a third-party site with fairly sophisticated functionalities.
That costs us nothing, y’all can print whatever it is that you want to print (or .pdf whatever you want to .pdf), all is well.
Episode 13 – The homepage shuffle.
Today’s the 1st of January, 2022. Which means I’m going to change the post date of the 30-ish articles using the new format to today.
This provides me with a simple means of knowing which parts of the site have been upgraded. If it’s dated 2022+, then it’s to the 2022+ format.
So there will be some shuffle on the homage, as the recently modified posts will appear in a different order.
If you’re a devoted WORG reader, the early stages of this work may not be thrilling – since I’m largely working on recent entries. These are easier to work with, but you likely already read them. So I’m also throwing some older entries in – which, for the vast majority of readers, might just as well be new.
The texts on the homepage will no longer announce that an entry has been upgraded. If it’s gone up the timeline, then it’s been upgraded.
Episode 14 – Let loose the poodles of strife.
For a while I only worked with a 15-to-20 set of entries, to design procedures and experiment with the new layout. But the deployment is now picking speed, to get a sense of feasible workload.
The majority of the entries being migrated to the new version are recent entries – reverse chronological order. These are easier to work with (more recent texts, more recent images, etc.) and I’m getting them out of the way whilst everyone is busy with new years transitions.
For people who regularly read the site, I’m throwing in about ¼ (aiming for ⅓ now) of older entries so it doesn’t all feel like reruns.
For the vast majority of readers, older entries redone to 2022+ standards are indistinguishable from new entries. Unless they just happen to have read them in recent weeks (say, after hitting our Random page).
Episode 15 – HREF hygiene.
A tiny thing. In the ported-over-to-the-new-layout entries, the links toward external sites now open a/ in a new tab (that’s more convenient) and b/ without passing information.
“Without passing information” is mostly for edge cases. For instance, if I once linked to a RPG blog, which then died, and had its domain bought by shady people doing shady things.
(Note that I don’t do that for Amazon affiliate links. If you decide you want to buy stuff on Amazon whilst perusing WORG (say, collected issues on Comixology) then by golly spend aaalllll the time you need on Amazon without distraction. *Something* gotta pay our bills.)
The “new tabs without passing data” has been the best practice for a good while. But it’s cleaner for us to do it manually, and the WORG 2022+ project means that I’ll go over every single entry by hand anyway.
It’s thus a good example of how each new version/upgrade of writeups.org is used to best-practise-standardise our mess. Because WORG well precedes modern best practises, so the #content (code, images, metadata…) was done in a way that doesn’t play well with modern frameworks and standards.
(You should have seen the articles’ code in the early 2000s. It looked like it had been written by Lord Humungus).
The 2022+ update also sees stuff such as constraining the dimensions of in-article images, which were previously all over the place.
So I guess that, eventually, it’ll all be clean. Making evolutions easier, and diminishing the need for multi-years reworking pass because the material is just too chaotic due to age.
Episode 16 – Better images (1/3).
We’ve used lightboxesWhen clicking on an image displays it at size, “over” the page before. Primarily in video games entries, when we wanted to show big screenshots with lots of details.
In pages adapted to the new layout, almost everything is lightboxed. As indicated by the magnifying glass icon.
Here’s an example, staring Poundcakes :
This solves several issues :
- Large pictures are part of our visual identity… but most readers see the site on a tiny screen.
- We want to display high-resolution images… but many readers are on slower, patchier network connections.
- Our images are compressed to a rate that was fairly generous… six or seven years ago. A “high definition” .jpg is not compressed at 51 anymore.
- Some of our images were just too tall for smartphone screens.
Lightboxing is our way of having our cake and eating it.
The in-article images (except for the headers) are in okay size and quality, and relatively lightweight. On smartphone they’ll look small, but everything looks small on smartphones.
For slower connections, the not-zoomed images are actually faster to load than they were before. Because they now have a maximum height. And also because we’re now using a better lazy loading solution.
*If* you want to examine the pic, then you can click and load a significantly higher-res, and often markedly larger, version of the image. If your screen is tiny, you can zoom and out – and move across the image.
(This can be useful if, say, you do cosplay and need to examine details).
Now, adding these amenities to the articles involves a lot of work. We do have a bank of large, lossless “master copy” images. But these are but a minority of the site’s images, and a percentage of these “master copies” aren’t good enough anymore.
So there’s a lot of images to be re-procured in sufficiently good quality. Say, redoing genuinely-high-def screenshots, or working with genuinely-high-def digital reissues of comics. That’s the largest chunk of the WORG 2022+ workload.
Episode 17 – Toward the ides of September.
Everything that was published in October, November and December of 2021 has now been brought to the 2022+ standard.
I’m doing ⅔ recent entries and ⅓ older entries. The idea was to get all the recent-ish stuff done in early January. Whilst people were still busy with New year workloads, recovering from any festivities, dying of COVID, etc.. To cut down on the feeling that it all was reruns.
I’ll mostly stick with the ⅔ – ⅓ mix. So for people who pay enough attention to writeups.org that entries from three months ago still feel too recent, we should soon reach August of 2021.
One entry — Bob the Goon — was done slightly out of sequence, to give me time to experiment with 2160p Blu-Ray footage for screencaps. Given enough processing, the final quality is fairly good.
But of course, many of the movies we like and have character profiles about… aren’t the sort of films that get 2160p remastered rereleases.
Episode 18 – Seeking search.
(This post was written during a 2-day span where the search function was nearly disabled. It’s now back up.)
Episode 19 – Better images (2/3).
There are many cases where the images we need to illustrate an article aren’t going to be great.
For instance, the only affordable copies of a comic book are microfiches or old scans. Or the movie was only ever reissued in DVD, with terrible resolution and encoding.
(For some reason, for years I thought that DVDs were 1080p. But no ! They’re 720×576 in MPEG-2. I blame inane marketing terminology around “HD”, of course – certainly not my failing memory).
That always was a problem, but it’s even more of an issue now that we want to present larger, lightboxed versions of the pictures.
For more than 15 years, I used a variety of techniques on Photoshop to do targeted blurring (or sharpening), denoising (Imagenomic filters saw a lot of use during the 2010s) and minor art reconstruction. These have become less necessary with comics being digitally reissued… though sometimes, the digital reissue isn’t too convincing.
That, plus the usual about background removal, colours correction, releveling, correcting printing/colouring glitches, etc.. And my usual failure point, making the word balloons look over-sharpened in order to get the rest of the picture right.
Much more recently, we’re gaining access to AI tools. I don’t use full-blown ANN tools à la ESRGAN, but the workflow can now tap :
- AI Denoisers (from Topaz Lab).
- AI .JPG artefact removers (from Adobe – those work great with blacklines art).
- Super-resolution NN zooming solutions.
It can often provide images that look markedly better than the available original does. So I guess that in many cases, our lightboxed images are the best version of the pic you’ll find.
(The main benefit from having spent hours polishing really good versions of images is that they get pillaged by people reposting them to make money and/or clout.)
Of course, almost nobody’ll stop and gasp. They’ll vaguely assume that the images we have access to simply were that good. OMG NOOO THEY SOOOO WEREN’T THIS STUFF IS AMAZING AND NOBODY NOTICES !!1!
Episode 20 – Categories in the craddle.
For several weeks, the links toward the categories and subcategories in the breadcrumbs were broken.
(Those are the things that say, say, Comics –> Marvel Universe atop the profile for a Marvel Universe character.)
The breakage was part of the switch in search solutions. Now it’s fixed. Which is nice, especially since — as it turns out — people are using those.
The FAQ about the search page has also been updated to reflect the new tech.
Episode 21 – Everything counts.
I use TinyJPG for compression on the *small* versions of images – the non-lightboxed ones.
It’s certainly not as lossless as they insist, but it’s a well-regarded solution to get lighter images with little loss of visual quality. Especially on sparse pictures, such as a small figure without background. So in-article images display faster, and you can lightbox them for a low-compression version.
The free WordPress version for TinyJPG has a 500 images per month limit, which I hit yesterday. It’s no big deal — I can do more on their site — but it means that 1,000 images have been redone so far in January.
(500 tinyJPG-ed small versions, and 500 not tinyJPG-ed lightbox versions, see ?).
It’s a more aggressive rate than I was expecting, but :
- I’m mostly working on published-within-the-last-six-months entries so far. Which can be processed faster as they need fewer improvements.
- There were two articles with a way-above-average number of illustrations. The weaponry-in-stories articles are notable for that.
Early on, the norm was one dirty, grainy, tiny JPG per entry. In more… recent years we’ve been cruising at about 6 pictures per average article, so we were about reaching 37,000 pictures on the site.
With the new inset+lightboxed logic, we’re of course mechanically headed toward 80,000. Twice the previous number, plus new stuff. But it’s a different approach, so it doesn’t quite count.
Episode 22 – Top ten.
Back in 2016 we experimented with a handful of “top ten” lists, since people *do* click on those. These were in a sidebar on the homepage, and in the footer – so they took little space.
The 2022+ version of the site has a revised version of it. The new Top 10 page has a pointer on the homepage, and in the footer. So that stays fairly discreet.
It still provides ideas for people who have no idea what to click on. Each year, one contributor suggests ten entries they like – and explains why.
Episode 23 – Better images (3/3).
When an entry is brought to WORG 2022+ standards, there’s a lot of small improvements. Especially if it was missing one copy-editing pass, or some pictures were still low-res, and such.
But for recent-ish entries, the bulk of the work is about images. Migrating them from the right-hand column, improving them, and having them conform to a now far more standardised size.
The last part is interesting. Normally, images used in website layouts can be cropped without much hindrance, or were produced to fit a set format (say, a cinema screen). But on WORG, the images are usually redone to be centered on the characters, and show as much of them (and their powers, their skills, their weapons…) as possible.
So in effect, the images have their own logic as to their format. Some will be tall and narrow, some will be broad but not tall, some will be square-ish…
Using a lightbox is a way to escape this constraint. Once it is displayed in a lightbox, an image can have whatever aspect ratio. It doesn’t have to fit in anything – not even the screen, as it becomes zoomable and explorable.
But the image that displays “at rest” — the not-lightboxed version — has to conform to a width, and to a maximum height.
- For some images, that’s not an issue.
- For most background-less images, doing a small version of the entire picture works well. And you see what you’ll get when you zoom in.
- But for some images, I’m stuck only showing a *part* of the shot. Here’s an example :
So in these cases clicking doesn’t show the same thing but bigger. Instead, it exposes more of the pic.
Which isn’t the standard behaviour on websites. Which means that most people will likely randomly discover it one day, and get more than they expected (but in a good way).
Another change is that most entries now have more images. Previously, I didn’t usually put images in the game stats sections, reasoning that most readers didn’t really read those. But nowadays, there’s enough images to decorate the game stats.
I suspect that this too will also mostly be discovered by accident. But yeah, in the new layout the stats are treated more like a part of the article and less like an appendix.
Episode 24 – Micro-moolah.
There’s one important aspect of the WORG 2022+ work you prolly don’t care about.
The material that was once in the right-hand column migrate into the #content proper. For shorter entries, this may mean that there’s too many elements for the article.
In a few cases, it’s images who get the axe. But they have to be genuinely bad or redundant for that to happen.
So more commonly, it’s the adverts that get thrown out. One Adsense ad gets deleted in every single case. And in most articles, the programmatic Amazon ad gets axed. So that’s two out of five gone.
Therefore, our ads micro-revenue is gonna drop, no doubt about it. Right as small costs overrun from the WORG 2022 contractors cost have us slightly in the red.
All because I’m an idiot who wants well laid-out articles with few interruptions and (when feasible) high-quality pictures.
There’s only one solution to this in the sub-lunar world. You got to the donation page and set a $5-a-month donation — or $10 if you’re obscenely wealthy.
As every destitute content creator on the Internet is bound to tell you, $5 is the price of one cup of that pumpkin spice latte thing people seem to like. I don’t know whether it’s good, I can’t afford one.