Tag: TODO: Blog Idea

Rats

Or maybe mice, or squirrels.

In any case, you don’t want them in your car, chewing on the wiring.

#

Fortunately they only actually chewed through the control wires for the high beams. Still going to be expensive to fix (and of course acts of rodent aren’t covered by warranty).

Going to need to inspect the garage, though, in case that’s where they came from.

—

We spent a chunk of the afternoon cleaning out the garage. Apparently something got into our backup emergency food and water and was trying to make nests out of bits of old stuffed animals. And silica gel. And old unbuilt model kits.

Katie’s going to do a second round of bleaching on the floor while I go pick up the car. And tomorrow we tackle the upper storage compartment, just in case the reason we didn’t find any actual rodents is that they’re up there.

And then get the bikes and everything that wasn’t contaminated or can be disinfected back in before it rains this weekend.

A lot of cloud-based, client-server applications really should be client applications with some kind of sync solution.I don…

A lot of cloud-based, client-server applications really should be client applications with some kind of sync solution.

I don’t need my shopping list to be reachable at any time from anywhere in the world. I need it to be on a handful of phones and maybe a laptop/desktop, and for changes on one device to show up on the others. That can be done with a web-based application, or a mobile app with its own backend. It can send updates halfway across the country to a dedicated central server so they can come back and reach the family member who’s at the store right now…

…but it can also be done by syncing changes peer-to-peer, or via a cloud-based relay, or building on a more general sync service like Dropbox or Nextcloud.

The web and cloud services have made the client-server model really easy to build for. As long as you know the client is always going to have a good network connection.

(Flashback to grocery shopping before they installed a new cell tower near the supermarket and we couldn’t even text each other “hey, I just remembered we need ___!” unless the one shopping was at the front of the store where the signal was just strong enough that an SMS message might show up. Obviously this would delay syncing too, but it’s another reason to keep a copy of the current data locally on the device.)

#

A related example where client-server makes more sense: actual shopping (whether arranging store pickup or buying something to be shipped). You need to know what’s available, what’s in stock, what the prices are…and so do all the store’s other customers.

The model for how people use it is a star: one central point (the store) and a bunch of rays connecting to it (the customers) and not to each other – so it matches a client-server structure.

Compared to the shopping list, where it might be built as a star, with the central server running it and lots of people connecting to it, but the actual use model is a small mesh: A handful of people and their devices sharing something among themselves.

Huh. the Arc browser is actually implementing *small AI features that make things useful*, like better title abbreviations in…

Huh. the Arc browser is actually implementing small AI features that make things useful, like better title abbreviations in tabs, choosing a better name for that download file that’s just IMG_12378612.jpg, creating hovercard previews for links, etc.

https://www.theverge.com/2023/10/3/23898907/arc-max-ai-browser-mac-ios

Key quote:

“People don’t want more internet. They want less internet.” If Arc’s AI tools were going to be actually useful, they had to help people spend less time dealing with their browser because, ultimately, most people would happily trade a delightful discovery engine for a few minutes of their life back.

#

Unfortunately all the AI features (“Arc Max”) are offloaded to OpenAI, except one that’s offloaded to Anthropic.

Ideally, I’d want anything that’s acting entirely on the content of the page I’m looking at (like the tab titles and filenames) to be done locally. Even the link preview should only really need to grab and summarize the remote page.

I don’t think getting a slightly-better summary on pages that don’t provide their own in metadata is worth the overhead of calling out to a remote AI service.

There are two main types of cell phone buyers

Intriguing insight in this review of the Fairphone 5:

There are two main types of cell phone buyers – those who love checking out the new gadgets and getting a new phone, and those who “hate saying goodbye” to their old devices, who would rather not replace a phone for mere planned obsolescence because it still works *just fine* except for this one little thing that either can be worked around or *should* be fixable.

Fairphone is intended for the second group

https://www.theverge.com/23895548/fairphone-5-review-price-features

On Wandering.shop

I’d love to get a Fairphone for this reason. But I also don’t want to get a new phone right now, for the same reason.

Here’s hoping what I’ve got lasts long enough for this model (or the next) to reach the US.

On Wandering.shop

Thinking about some basic privacy/offline principles for computer development

  1. If something can be done entirely locally on the user’s own device…it should be built that way!
  2. If something needs outside information, but it can be collected passively (ex. GPS location), it should be built so that it can.
  3. If something needs to interact with another system to do what it’s supposed to do, it should only connect with those that it needs to.
  4. Corollary to that: trusted peer-to-peer is usually preferable to over client-server-client. A trusted server may be better than untrusted peer-to-peer.
  5. If something needs to interact with another system to do what it’s supposed to do, it should only send information that’s needed.
  6. Any system should only keep the data needed to do its job, and possibly for troubleshooting.

Some examples:

Firefox’s local get-the-gist-of-a-webpage translation vs. Google’s send-the-text-to-Google-Translate.

OSMAnd can download a regional map ahead of time and do all the navigation routing on the device without a network connection. (You can’t get live traffic that way, though – that’s something that does require a network connection.)

eBook readers usually have no problem letting you read a book offline once you’ve downloaded it. The same should be true of text files, PDFs, email, locally stored music and video, RSS articles, etc.

Anything that is available over a LAN should be reachable even if there’s no remote connection to the internet.

Speech recognition should be done entirely locally.

I should be able to sync my laptop or tablet or phone, then pop onto a boat or a plane or into a diving bell or onto an underground train or go out into the middle of nowhere for a weeklong camping trip, or just turn off the network…and as long as the device still has power, I should still be able to read old emails, write new ones (and queue them up to go out when I get back to a connection), read a book, use the map, read articles I’ve saved up to read, take photos, review photos, delete the ones that didn’t come out well, crop or adjust the ones that need something extra, play a multiplayer game with my kid on two devices in the same room, write a draft of something…

The old always-offline and the new always-online are not the only design models available, and they’re certainly not the only situations people find themselves in. Just imagine tethering your laptop to your phone in an area with spotty connection. There are places and devices where I can barely get the Nextcloud login screen to render. And that’s my own server.

The idea that everything is going to have a constant internet connection makes surveillance tech even worse, because

  1. It’s easy to offload processing to your server even when the phones can handle it.
  2. It’s easy to build in things like update checks and news.
  3. Once you’re already doing that, why not pass a little more info for analytics or targeting.
  4. If they’re always online, you don’t need to wait for them to open it up, you can pop up a notification to grab their attention.
  5. If they’re always online, you can collect data more simply. You don’t need to wait for a connection, you don’t need to queue up multiple batches of telemetry, you can just send it.

TL;DR:

Run locally, sync remotely.
Only sync what the user needs you to.

#

Does The Instance You Pick Matter?

Mostly it doesn’t matter for the person using it*, so you can just pick one that isn’t overloaded to start. But…

Ways it does matter:

* Your instance’s moderation policy and actions. (including what content is allowed/disallowed, how they deal with harassment, etc.)
* Server reliability. This can change drastically if a lot of people join at once, as many Lemmy sites have discovered this week! (I believe Lemmy.ml and Lemmy.world have both upgraded their hardware in the last few days to deal with this!)
* Admin reliability. This is harder to tell up front, but it’s worth taking a quick look at whether the admins seem to be active and responsive, whether they seem like they’re in it for the long haul or if they’re experimenting, etc.
* Aaaand I just read about the situation with Beehaw.org defederating from lemmy.world because their mods were overwhelmed, so that (for now) the two servers can’t interact with each other.

Switching is sort of easy in that all you have to do is create a new account somewhere, and you don’t need to tell your followers because Lemmy doesn’t have user subscriptions (though someone could follow you from, say, Mastodon)…

…but it’s also not easy in that Lemmy doesn’t have tools to export/import your subscriptions (yet?) so you have to add them to the new account manually. And moving your posting/comment history isn’t something that’s doable at the moment, either.

What I did when moving from lemmy.ml to lemmy.world was put the old/new accounts in each others’ bios and add “Old Account” to the old one’s display name. I’m not too attached to my post history sticking to my profile.

*I think it matters a bit more for where you set up a community, on the basis that an instance focused around, say, history would be a better place to create an archaeology community than one focused around FOSS. Though you might want to cross-post articles about free software used in archaeology!

[Discussion thread deleted]

Nextcloud Apps

If you use a Nextcloud server, there’s a good collection of apps (some official, some third-party) that work with it. The ones I use:

Nextcloud – main app, does authentication, file access, optional auto-upload photos
Nextcloud Notes – kind of like Google Keep, but simpler. (IIRC Carnet is more like Keep, and also open) Nextcloud Talk – instant messaging, supposedly can do voice but I’ve never used it for that
Nextcloud News – RSS reader that syncs your feeds and read/unread through your Nextcloud server

Plus these apps that aren’t Nextcloud-specific, but work with it and other sync methods:

OpenTasks – ToDo list (needs Dav5x to sync)
DAVx5 – Syncs contacts, calendars and to-do items between any CalDav, CardDav or WebDav servers and your Android system, so you can access them with any local contacts or calendar app. (For instance: K-9 Mail can use contacts from my Google account and my Nextcloud account, and Simple Calendar can do the same with my calendars.)
Floccus – Bookmarks manager that can sync across multiple desktop browsers and the mobile app, using any of several sync options including Nextcloud

#

Flash TV finale

I haven’t watched it, but apparently the big villain in the 4-parter is…

Cobalt Blue.

CBR describes him as a “classic Flash villain” and “iconic.”
https://www.cbr.com/the-flash-cobalt-blue-set-photos/

I’ve been out of the fandom for a while, but back in the late 90s early 2000s there was a large contingent of Flash fans (specifically Barry Allen fans) who HATED the character. Mostly because in the comics he was Barry’s separated-at-birth twin.

On Wandering.shop

I find this hilarious. Even if it’s just the fact that we’re still in the 1990s phase of the nostalgia cycle, the idea that Cobalt Blue is worth using for a TV series finale? Never would have occurred to people back in the day.

And TBH I kind of feel vindicated for the blog post I wrote back in 2008 arguing that if the character had been introduced back in the Silver Age, he *would* have been a classic.

Cobalt Blue, Classic Rogue?

I guess now he is!?

On Wandering.shop

Kind of wish I hadn’t deleted that CobaltBluePrime handle on Twitter back in November. It would’ve been perfect to bring it back and post “AT LAST MY TIME HAS COME” or something like that.

(I set it up for a 2011 April Fools’ joke where the blog was rebranded as Cobalt Blue Online for the day and we posted a bunch of Cobalt Blue-themed articles)

On Wandering.shop

Though I suppose I shouldn’t be surprised. I mean, this is the show that successfully transformed Vibe into a beloved character!

On Wandering.shop

Re: Juice Jacking

Replying to a post about this alert warning against using public phone chargers in case they were modified to install malware.

The other day I was wondering how effective the default “no data transfer” mode on Android is at protecting against this type of attack. Unfortunately the articles I found tend toward either “malware could get around that too! Even if your phone is off!” or “this has yet to be seen outside security conference demos.” So, not terribly conclusive!

On Wandering.shop

re: Congressional TikTok hearings

Rep. Weber just accused tiktok of pushing chinese propaganda and “woke” propaganda in the same sentence — thedextriarchy

I think their mind would be blown if they knew what Chinese state ideology thinks of the things they call “woke.”

Not exactly big on minority rights, racial equality, acceptance of non-traditional gender roles, or awareness of and wanting to dismantle state abuses of power in enforcing conformality.

On Wandering.shop

No, brain, it’s Martin Guerre, not Martian Guerre#sleepy

No, brain, it’s Martin Guerre, not Martian Guerre

#sleepy

On Wandering.shop

OMG it would be a perfect match: Astronaut goes to Mars, comes back…different. Is it really him? Is an alien presence hitchhiking in his mind? Or is it a shape-changer like The Thing? Or did he simply have a life-changing experience on the mission?

It could be played for comedy, horror (body or psychological), or anything in between.

On Wandering.shop

Side project: improve Fediverse interoperability

I’ve made it a side project to improve #Fediverse #interoperability. Currently:

Using multiple platforms of different types:

  • @Mastodon (microblogging/general)
  • @bookwyrm (book reviews and library tracking)
  • @LemmyDev (link sharing/discussion forum)
  • @pixelfed (photos)
  • WordPress+ActivityPub plugin (blogging)

Testing other platforms that fit the same niche as Mastodon:

Finding what does and doesn’t work between them.

Reporting bugs and testcases.

So far it’s been mostly haphazard. I need to start tracking what I have and haven’t tested.

And I’d like to make time to fix some bugs and maybe implement some features too — I just need to carve out the time to dig into some code!

#

I know there are more ActivityPub platforms out there, and I’m sure I’ll try out more of them as time goes on.

For now, though, I don’t want to spread myself too thin on the general platforms, and I don’t have a good personal use case for, say, Funkwhale or Peertube that will make sure I actually keep using it!

#

Is there a general #federation #compatibility matrix somewhere? I would have thought FediDB or SocialHub would be the place to look, but I don’t see anything of the sort on either site.

#

Great, I’m already thinking about setting up a Codeberg repo with just issues and docs linking to each projects’ federation-tagged issues.

C’mon, brain, save it for after work.

EV car batteries that no longer hold a strong enough charge for driving can still hold a strong enough charge for grid storage

I remember reading about this idea a while back: EV car batteries that no longer hold a strong enough charge for driving can still hold a strong enough charge for grid storage, fitting the “reuse” part of reduce, reuse, recycle.

The big reuse: 25 MWh of ex-car batteries go on the grid in California

On Wandering.shop

uspol/scotus: How weird is it…

How weird is it for
(a) Gorsuch to be the swing vote and
(b) Gorsuch and Jackson to co-sign a dissenting opinion?

(Basically, whether you think you need to summarily kick people out at the border or not, you can’t justify it as keeping covid out of the country.)

No surprise, the rest of the “conservative” majority voted to force the government to keep doing it, while the liberal judges agreed that the policy should be allowed to expire.

https://wapo.st/3GrwTBj

On Wandering.shop

visual #migraine, weird

I’m fortunate not to get the actual headaches, but I have other #migraine symptoms on a regular basis, including the visual distortions. For me it’s usually an open curve, with the line filled with zigzags flashing different colors, that starts at the center of my field of vision and moves to the periphery.

Last night I had one start *during a dream* and persist when I woke up. The weird part: the curve was rotated 90°: Open downward in the dream and leftward when awake.

On Wandering.shop

Also I think it might have been grayscale in the dream (with full-color perception of the rest of the dream) and color when I woke up, but I’m not 100% certain

On Wandering.shop

A link shortener is just an extra layer of complexity that can go wrong.

Reply to a poll about custom link shorteners

Personally I think a shortener is just an extra layer of complexity that can go wrong.

If you’re in control of your own URL scheme for your site you can make it human-readable and short enough to avoid word-wrap on email, and it’s not going to take up any extra space on Twitter, Mastodon or anything else unless you’re sending it in an actual SMS message.

Who Are Les Miserables?

A response to @matthewjmandel asking my thoughts on A Comparative Book / Movie Review of LES MISÉRABLES

It’s interesting. I agree with a lot of the comments about losing complexity, but I don’t have as much of a problem with the character changes (partly because I’m used to the stage version, where Gavroche is less political & the the Thenardiers are funny, but still dangerous)

Eponine’s probably the biggest change that isn’t just a simplification, but I think her role in the story still works, even if the details have been changed.

I do have a problem with the finale, because it’s *not* Jean Valjean’s heaven by any stretch of the imagination. It works better on stage, where it’s more like a curtain call for all the characters who have died.

The main place I disagree with the post, though, is about the theme and title. Listening to @readlesmispod talking about how the word is perceived in French makes it clear that *all* of the main characters are “miserables” and Hugo is linking the sympathetic wretched like Valjean and Fantine with the clearly evil wretched like the Thenardiers because, as far as society is concerned, they’re the same. Society looks at Fantine and thinks she’s just as depraved as Thenardier.

And Hugo is arguing that they *all* deserve compassion, that they *all* should have a better life, that society should treat them *all* better, whether they turn to evil when they fall or not.

So the musical is less of a complete inversion of the theme and (once again) more of a simplification.

Thinking about different aspects of location tracking in terms of…

Thinking about different aspects of location tracking in terms of

1. Your device figuring out where it is.
2. Others figuring out where your device is.
3. Your device telling others where it is.

1 would include GPS, or just listening for cell/wifi signals.
2 would include IP geolocation, or seeing which towers your calls go through, or tracking which map tiles you download while moving around.
3 would include telling a website, or an app uploading your location

1/

On Wandering.shop

Pinging hotspots or cell towers is both 1 and 2, figuring out your own location while revealing clues.

So you’ve sort of got passive sonar, active sonar, communications metadata, and actual reporting.

As for tracking:
Passive sonar isn’t a risk
Active sonar’s risk is inherent to being active
Metadata’s risk is inherent to actually using a connection
You can obfuscate it by things like onion routing, vpns, extra requests, pulling location-specific info ahead of time, etc.

2/

On Wandering.shop

And then there’s the issue of when your device explicity sends your location once it’s figured it out. Like when you search for the nearest coffee place or bank branch, or you’re using navigation software, or updating an online map in real time. Or when an app or website just wants the data.

Fortunately, OSes and browsers have decent access controls for letting apps and websites know your location (as determined by your device). But an app or website can still try to infer from other info

3/

On Wandering.shop

An app with online ads could look at which CDN endpoint you end up connecting to. Or make connections to sites with known locations and see which is faster, just like a speed test app or site choosing the nearest server to make a proper test. It’s a lot less precise, but it’s not like the OS can block the upload as easily as it can just refuse to hand detailed location to the app.

And of course it can be correlated with other metadata like geolocation

4/

On Wandering.shop

And of course anything you *intentionally* report — online navigation, check-ins, Pokestops you spin, the nearest ATMs on the bank site — you have to trust that the service at the other end isn’t going to misuse it.

Your bank isn’t going to care. They already know which branches and ATMs you visit.

I wouldn’t trust an ad network with a 10-foot pole.

Something big like Google or Microsoft or Amazon? That’s where you get into trade-offs.

5/

On Wandering.shop

Offline maps & navigation definitely have an appeal to them. Especially if you aren’t sure you’ll have reliable network access where you’re going. It’s doable. The only downside is you can’t add live traffic info. You could probably download typical traffic per day/hour along with the map and get rough esimates, butof course not be able to know there’s a 10-mile slowdown because a truck jacknifed, fell over, spilled an entire load of cheese and caught on fire

6/

On Wandering.shop

or maybe grab traffic data for a wide area, quantized so that it only reveals which large rectangles you’re passig through, not which roads. Like iNaturalist does with obscuring locations of observations.

Anyway, I don’t know why I’m still writing this. I just wanted to write down the active/passive sonar analogy, and i kind of rambled on from there.

7/7

On Wandering.shop