Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:00
It's time for the quarterly dusting.
0:02
Wait, you only dust every quarter. I've
0:04
certainly never drugged my computer into the
0:06
PC world office to shoot a video
0:09
and then put it up on the table. And Adam
0:12
went, what the hell is wrong with your computer? And
0:15
then, and then suddenly you were shooting a very different kind of
0:17
video. Look, I was having some crashing
0:19
problems. It wasn't the dust's fault, but I figured
0:21
I should leave it there just in case, but
0:24
I have to dust way more than once. Like, like
0:26
I, before I moved to California, I'd never lived someplace
0:28
with hardwood and I'd never lived someplace
0:30
that has, that doesn't have
0:32
central, like central air that
0:35
has some sort of centralized filter that sucks a
0:37
bunch of dust and particulate out of your air. Oh,
0:39
that's why it's so dusty around here. Yeah. Well,
0:41
plus we live places where there's a lot of cars.
0:43
So like urban places are dusty because of all the
0:45
cars, which is, I'm sure great for us
0:48
too. And we live close to the
0:50
water. I don't know if that air circulation would make
0:52
it more or less dusty. Well, the fog, it turns
0:54
out, picks up a lot of particulate. And then when
0:56
the fog lands on things and condenses it, yeah, that's
0:58
mostly outside though. That's an ugly word.
1:01
Particulate. Particulate's bad. It turns out
1:03
usually. Yeah. Yeah. I don't like it. Also, I
1:06
really says everything about why you and I do
1:08
this podcast that when I talked about dusting quarterly,
1:10
you went straight to computers. You didn't even, Oh,
1:12
I knew what I look, didn't even pause to
1:14
consider. I was talking about household. You can't, I've
1:17
lived here for long enough to know, and I know
1:19
you well enough to know that there's no way you
1:21
only dust your apartment once a quarter. No, I cannot
1:23
stand it. I'd ideally a once a week. Yeah. It's,
1:25
it is a, it is one of the great things
1:27
about having kids is that now that's the kid's job.
1:29
She has to go around and does stuff. So that
1:31
means all the, well, okay. So
1:34
the thing that became clear with this new
1:36
regime is that we're three different
1:38
heights, right? Like I'm tall, my wife's the
1:41
mid-size and my daughter's the short one. And
1:43
all anybody ever does is stuff at
1:45
their eye level and
1:48
below. Sure. The internationally recognized standard for dusting. Yeah.
1:50
So like, when you look at the top shelf
1:52
and our living in our dining, you know, we
1:54
have the shelves up against the dining room wall.
1:56
There's like some stuff up at the top. It's
1:58
not good up there. It's that's a
2:01
that's a I have to get a step ladder
2:03
and get in that I get it. You know,
2:05
even even I have a bit of an out
2:07
of fight out of mind rule about dusting the
2:09
top of the microwave, which is on a like
2:11
a shelving unit. Doesn't get dusted very say our
2:14
our microwave is above the stove. It's one of
2:16
those ones that hangs up on the wall above
2:18
above the stove. And it is
2:20
a real double whammy of horrors because
2:22
it gets like grease splatter from the
2:24
stove from the range top. Like
2:27
the smoky grease. It just oh yeah,
2:29
things and gets on cabinets and stuff.
2:31
Kicky, greasy dust just. Yeah,
2:33
then the dust gets in there. It's like
2:35
a horrible amalgam of of anyway.
2:38
So as the person who
2:40
invented, I believe the hey, if
2:42
you take a swiffer and tape it over the fan
2:44
and takes on your computer, the inside of your case
2:46
will be a lot less dusty. Maybe
2:48
20 years ago now. Wait, yes. Did
2:51
you actually propagate that idea at a time when
2:54
nobody else was saying that in a magazine and
2:56
it became very popular? Wait, really? Yeah. Wow.
2:59
I swiffers came out. They were they just introduce
3:01
swiffers. I was like, oh, shit. Yeah,
3:03
this is because I've been using dryer filters before that,
3:05
but they're not as good. The Swiffer is
3:07
incredible for pulling dust out of the air. The
3:09
modern cases have fans and filters and the whole
3:11
thing. So the light lets it lets of an
3:13
issue. But back then this was pre who's built
3:15
in filter on your on your fan intake. This
3:17
would have been in an early 2000s maximum PC,
3:21
maybe even a late 90s one when I was still a freelancer.
3:23
I put it in the doctor column, if I recall. I
3:26
haven't done the science. I don't know that that was the first recommendation
3:28
for that, but it was the first time I had ever heard of
3:30
it. And I came up with it and people
3:32
at the office were like, what are you doing? And I said, well,
3:34
I put it there and then the dust doesn't go in the computer.
3:36
And they're like, wait, what? You know, even
3:38
with modern filters, it's a little bit of dust
3:41
in there. Use a little bit better Swiffer. So
3:43
what how's your dusting going? Is it I guess
3:45
the question? It's not OK. Let's say the NASA
3:47
has been running continuously since September. Yeah. Without
3:50
being touched. I bet that's pretty nasty. Probably,
3:54
probably to just it. Oh, I'm going
3:56
to pull it out. Oh, dude. Oh, getting that
3:58
thing out is a production. Oh,
4:00
is it like wedged in the corner between your desk and the
4:02
shelves or something? It is wedged as tightly
4:04
into a tiny space that is exactly the size
4:06
of that, the SLATX case as humanly possible. And
4:08
I have to like move this full upright keyboard
4:10
out of the way. I have to move furniture
4:13
to get the mass out. Yeah. Once
4:15
upon a time, I had a computer... There
4:18
was a period of time when people were making, I
4:21
think Zalman maybe made a case that had like two
4:23
cases in one big square and you
4:26
could put like one computer on one side, one computer on the other. I
4:29
put my server in one side and put my
4:31
desktop in the other and then it's just one
4:33
big case. And I had that
4:35
for about 45 minutes because like once I built everything
4:37
up in there and I put it back in the
4:39
corner and I was like, oh crap, I didn't get
4:41
a cable and I had to take it all the
4:44
way back out and it weighed about 80 pounds. Oh,
4:46
it's a nightmare. I... Don't
4:48
get me wrong. Yeah, no, I took it back to
4:50
the office and went back to my old cases as
4:52
soon as I pulled it out from back in the
4:54
wall. Yeah. Yeah. Feel your pain.
4:57
Always the trade off. Always making them tidy
4:59
versus maintenance. You know, like the more tidy
5:01
it is, the harder to maintain. So I'll
5:04
say I bought that plug-in canned air thing,
5:06
like the plug-in blower and it's
5:08
been maybe the best thing I've ever bought. Oh,
5:11
great. Because when I go out, when like
5:13
my vacuum filter is clogged up and crappy,
5:15
I just go in the backyard, put on
5:17
a breath mask and let it rip. Just
5:19
blow that thing out. Those things are so
5:21
boringly great that like it's made discussing the
5:23
actual dusting very mundane because what is there
5:25
to say? I take the cases out on
5:27
the back porch now and blow them out
5:29
for 10 minutes and that's all there is to discuss.
5:32
Mine has a light on the front, which I quite
5:34
like. That was a good... That was... I
5:36
didn't mean to buy that, but I'm glad I did.
5:38
I also have to say we got a
5:40
new vacuum cleaner this year because our old vacuum
5:43
cleaner, the battery was dead in. It
5:45
also has a light, speaking of dust. And
5:49
when you're vacuuming the hardwood floors and
5:51
there's a low-level light shining, you can
5:53
see everything. Oh, no. It's
5:56
really good. There's about a 10-minute window every
5:58
day, late in the day. when the sunlight
6:00
hits just the right angle, you can see all of
6:02
the hair and dust and crud on
6:04
the hardwood. And it's like, better
6:07
off not knowing. Well, look, I like to use
6:09
the light so that the dust sees what's coming
6:11
before I destroy it. You know, it gets sucked
6:13
up in there and then, well,
6:16
you know, it's gone. Anyway, it's the
6:18
promise of modern technology. Strike fear in the heart
6:20
of dust. I was gonna say, my vacuum
6:24
is a transformer. Okay.
6:27
It's a three-way Miele. And
6:29
it can be like a handheld dust buster, ideally.
6:32
Like that's pretty standard if you're used to your Dyson
6:34
stick packs or whatever. But you
6:36
also can jam the heavy part down into
6:38
the beater and it'll sit up vertically
6:40
and just stand. It doesn't have to be hung from the
6:42
wall or anything. Or you can put
6:44
the heavy part on the end and have it be like a
6:46
normal stick pack. So it's a
6:49
triple changer is what you're saying? Yeah,
6:51
it's like a third-gen transformer. It's
6:53
not like you can buy four others and make a
6:55
super duper big vacuum. But you
6:59
can transform it three ways. Now
7:02
I'm picturing the Devastator of vacuum cleaners. We
7:05
should probably just start this podcast. Welcome
7:28
to Brad and Will Made a Tech Pod.
7:30
I'm Will. I'm Brad. I
7:32
am not gonna be able to go of the image of
7:35
a Gestalt vacuum cleaner, traction. I want nothing more. Yeah.
7:39
Very long time. In fact, could we just do that with
7:41
all household electronics? Could we just make, buy five of them
7:44
and plug them in the right way and they become a giant robot?
7:46
How, like what's, there's so many ways that
7:50
could go back and forth. I mean, I think it's
7:52
a very simple process. I mean, I think it's a
7:54
very simple process. There's so many
7:56
ways that could go bad though. I
7:59
want the Predacon. I like. Like the Constructicon, the
8:01
Devastator was pretty good, but the Predacon,
8:04
that was big and orange. It was a chunky
8:06
boy. Was that the Predons? I thought they were...
8:08
I think they were. No, no, the Rheomorrhizecticons.
8:11
The Predicons were like... The Rhinoceros. It
8:14
was like a Rhinoceros and a Panther, I think.
8:16
Sure. I can't remember the other ones. Maybe an
8:18
Elephant? I believe that. Several
8:20
animals that I did not think of as
8:22
murderous and then a couple of giant cats,
8:24
which clearly were murderous. You know, anything can
8:26
be murderous if you're enough of an evil
8:28
robot. The Insecticons were solid
8:31
though. What are we talking about this
8:33
week, Brad? What do we got on the podcast? On
8:35
the docket this week. A twofer. Oh,
8:37
I love a twofer. I think is how you
8:39
would put it. I do like a twofer. We've
8:43
been ongoing brainstorming around the show
8:45
topics, show formats, stuff like that.
8:48
As we said before, we like the
8:50
topic episodes. Yeah. But there are
8:52
a lot of topics that we'll kick around. In fact, I kind
8:54
of wish we had a list of every time we've done this
8:56
over the last four... Oh my
8:58
God. Almost five years. Yeah, there's a lot that we
9:00
get to and we're like, oh, this is it. We
9:03
can't do a full 45 minutes or an hour on
9:05
this. Yeah. But a number of times we've kind of had
9:07
a topic that's like, man, that would be a good 15, 20, 30 minutes maybe. But
9:11
that's not going to support a whole episode. Yeah. There's
9:14
a lot of those. Yeah. Anyway. So
9:16
we have two today. Yeah. I think
9:18
we might start doing a little bit more of this kind
9:21
of like split episode, like two topics of that nature that
9:23
can be discussed pretty thoroughly in like 20 minutes. So this
9:25
week, I've been making
9:27
a new benchmark, working
9:30
on a project for PC World. And one of the things
9:32
that we're working on is performance
9:35
for game developers. So
9:38
things that you use a computer
9:41
for that take a lot of time. As
9:43
I was working on this, it made me realize
9:45
that building benchmarks... In the
9:47
past when I built a benchmark, it's usually taken
9:49
something that's found like a video game and then figuring
9:51
out how to make a repeatable benchmark out of
9:53
that. In this case,
9:56
I'm actually setting up the environment and stuff
9:58
like that that I want to test. And it made
10:00
me realize it's kind of an inherently political
10:02
experience. We'll talk about it. Interesting. Just
10:04
not the term I was expecting to hear, but I'm curious
10:07
to hear more. And
10:09
then we're going to do kind of an,
10:11
I guess, an update on matter and thread,
10:13
the nascent kind
10:16
of universal home automation standard. Longtime
10:19
listeners should get out there, print out of the
10:22
Garnet, Troff of
10:24
Despair. What
10:27
is it? Yes. I can't remember
10:30
what that's called now. I can't believe I'm blanking. Troff
10:32
of Despair, I believe. Well, the Troff of Despair is
10:34
where we're at, but they call the product hype cycle.
10:36
Oh, the Gartner? Gartner product
10:38
hype cycle. Yes. Yes. That
10:41
one. Yeah. Yes.
10:45
But it's going to be okay. Don't worry. Yeah.
10:48
You know, we talked about home automation off and on on this podcast.
10:50
And I think it was the Fospod with
10:52
Paulus from Home Assistant, where we really kind
10:54
of went somewhat in depth on thread and
10:56
matter, which is supposed to like unify all
10:58
of the warring kingdoms of different home automation
11:00
standards. I mean, I would say that there's
11:03
probably an XKCD for this particular example. That
11:05
will be referenced later. Yes. Yeah.
11:08
But I've been thinking about home automation stuff again
11:10
as like some of my Wi-Fi and smart plugs
11:12
are not the best. And
11:15
I've kind of always felt a little irresponsible about
11:17
not putting all of those on a VLAN anyway.
11:20
And I've been thinking like, what if I just got out
11:22
of the Wi-Fi automation game and
11:24
looked at thread and matter or
11:26
maybe Zigbee and Z-Wave, the older
11:28
stand. Anyway, I've done a little
11:30
reading on how thread and matter have shaped
11:33
up or rather have not shaped up. You
11:35
sweet summer child. Yeah. So anyway,
11:37
let's get into it. Do you want to talk about benchmarking first
11:39
and then get to the whole... Yeah. Yeah.
11:44
I mean, it's such an underserved market. Like
11:46
how many benchmarks are being done for game
11:48
development practices out there? Everything, like you said,
11:50
everything is Cinebench or like video games. Well,
11:52
yeah. And so Puget does... Pugetbench
11:55
has a bunch of workflow tasks that are
11:57
common for workstation users. It's really good. But
12:01
not a whole lot of like unreal compiles partly
12:03
because in order to get a realistic benchmark you
12:05
need something That's probably a proprietary asset and
12:08
and honestly getting unreal up and working
12:10
in a in a Way
12:13
that it will build a level is
12:15
is fairly time-consuming. So I Looked
12:19
at UE 5.3 and said hey
12:21
if we just take one of the default
12:24
games that they ship and then jam some
12:26
landscape and geo in there That's
12:28
stuff that we buy assets from the store Then
12:32
you know, we can make something that
12:34
is representative but doesn't take You
12:37
know 14 hours, so I've been working on this it's one of the things I've
12:39
been working on when we're not doing on camera stuff at PC world over the
12:41
last few weeks and I shipped
12:44
the first version of it to some testers this week
12:47
But as I was doing the final as
12:49
so so just to understand what we're talking
12:52
about When you when you
12:54
ship a level in unreal at least Depending
12:57
on what lighting choices and stuff you use There's a
12:59
whole series of things that have to be built like
13:01
the the nav mesh has to be built and and
13:03
there's a bunch of computational stuff that has to happen
13:05
and the lighting bakes have to happen if you're using
13:07
pre-computed lighting and your
13:09
your LODs for the foliage have to be baked
13:12
in and and LODs
13:14
for everything have to be baked in All
13:16
and and you can basically hit a button
13:18
and unreal after you've opened up that level
13:20
that says build all levels and then start
13:22
a stopwatch and Time how long
13:24
that takes to run and if you're on a production
13:27
game like I'm the game that I worked on the
13:29
end of cruises Is an indie
13:31
game made by like 20 people at most? So
13:34
it was a little for a GPU light bake
13:36
on that. It was like 40 minute 20 to
13:38
40 minutes depending on the level We're
13:41
doing the same thing on the CPU light bake take a
13:43
few hours probably for each level eggs On
13:45
on what kind of CPU are we talking to do? What
13:48
anything? For 3900
13:50
K. Oh even a very good one even a very good
13:52
one I'm
13:55
gonna I apologize. I'll probably do this repeatedly. I'm
13:57
going to ask you to define a couple things
13:59
as well. Or or elaborate on
14:01
things. Would what is an admission is for me
14:03
some navigation measure that doesn't yet have passing for
14:06
a I guess. So it's the thing that tells
14:08
a basically minded I'm not the person asked by
14:10
the strom, give an answer and people yell be
14:12
in town i'm Bronx or and and comments on
14:14
i'm wrong but it's. It's
14:17
the thing. It's the underlying piece of
14:19
unreal technology. the Alps, the a I'd
14:21
pass from notes Notre Dame. So basically
14:23
it looks at all of the places
14:26
he marches, navigable in the game, at
14:28
all the places you said hey, You
14:30
can't go here and then connects them all
14:33
to each other's best they can and if
14:35
it doesn't find ways from to catch it'll
14:37
it'll for flags and even so you can
14:39
literally hit p I think it unreal and
14:42
it'll turn the floor green where the a
14:44
i can controversy because there's there's most evil
14:46
builds up on top of that like we
14:49
built our as the and presume to the
14:51
whole context layer. On top of that that
14:53
told the i like what was like wow
14:55
to find him. It
14:58
sped more informations yeah that was needed for the came
15:00
to. Work But sir, First, you do the admission than
15:02
you do the other stuff on top of them. Have
15:04
a sense. The other thing I was gonna ask
15:06
you kind of kicked us are saying the you
15:09
basically started with of the where you put it
15:11
was one of the default the games that they
15:13
ship for the duration of so I can do
15:15
to like what what does that What do they
15:17
quote Unquote game and compass and my contacts as
15:19
a just say is just a camera perspective and
15:21
set of basic mechanics or and like are there
15:23
other geometry goes around us ago without like what
15:25
you mean by game in the context of like
15:27
unreal project So it did it. Basically you eat
15:29
ships with million the version you have a bunch
15:32
of different default projects there some you he five
15:34
ships with some. That are like architecture
15:36
and and live some generation projects.
15:38
There's several game pamphlets like a
15:40
shooter this their shooter.the Xp which
15:42
is a first or third person
15:44
shooter. There's a third person shooter
15:46
exits shooter Idiocy The first horses
15:48
here is a third person shooter
15:50
template. There's a racing game template
15:52
I think us and basically it
15:55
it ships. It means that there's
15:57
a camera. There's awesome. Some sort
15:59
of be. The the blueprints of set
16:01
up the basic menus and the first level
16:03
of a game or second level a game.
16:05
Odd that there will be lighting and gun
16:07
mechanics in the case of the first person
16:09
shooter and has likes of of oh and
16:12
i'm he messes player mess by and also
16:14
has like the first person arms that used
16:16
to hold the gun, stuff like that and
16:18
animations to basically get they got so sick
16:20
of it as a tutorial. So.
16:22
Basically they have a lot of documentation the sites.
16:24
Okay first you can start with shooter id exceed
16:26
your video to this part of the project and
16:28
do this and this and this. and then you
16:30
can open up the simply prensa didn't do this
16:32
and this and this and bomb It's a it's
16:35
a jumping off like. I. Don't know
16:37
that everybody does as. A lot
16:39
of people start with seizure A
16:41
Cc. You're using the same basic
16:43
skeleton and rigs. For. Your
16:45
your player characters and your by paypal humanoid
16:47
characters are because then you can more easily
16:50
use asses off the sore and stuff like
16:52
that. their lot of Acid Caesar, same rigs
16:54
and skeletons. So I'm and it
16:57
is. There's real benefit to starting with. that
16:59
is my understanding that I'll make sense to
17:01
have any idea how much you would have
17:03
to differentiate from the whole project before you
17:05
can ship and per the terms of service
17:08
or whatever ice I assume you can literally
17:10
just mythical Be kind of funny if somebody
17:12
did just thought process so I mean I'm
17:14
sure he can. Ah, the default project. Usually
17:17
they don't compile out of the box such
17:19
as com you have to a buddy, but
17:21
like when you go through the whole project,
17:23
the whole tutorial series. as usually I twenty.
17:25
Hours long and Thirty hours long. You'll end
17:27
up with something the you can run in
17:30
the A fc on your desktop. I'm. I.
17:32
Don't think they would tear I don't know I
17:34
any be would do that. It's hims again of
17:37
a waste of of course it was his father
17:39
going to be financially lucrative with my for you
17:41
just avoid funny met a commentary movement but like
17:43
just him like the levels are like. It.
17:46
It's like ads basketball gymnasium
17:48
size room with no roofs.
17:51
and know wall like a great wall of
17:54
grey box walls grey box floors stays like
17:56
a couple of physics objects is a ramp
17:58
and of and that's instance And
18:00
then like a couple of pillars and some spawners that
18:02
just spit out enemies So it's like it's
18:04
not super exciting stuff But it is really nice because
18:07
it turns out when you're starting a project from scratch If
18:10
you if you haven't done that before just getting
18:13
the project up and running is pretty hard It's a
18:15
it's a it's a it's a challenging
18:17
endeavor So starting with that Bypasses
18:19
a lot of that stuff Unless you learn
18:21
about like menus when you need to learn about
18:23
menus instead of menus as the very first thing
18:25
you learn sure So
18:27
anyway, so I started with that The
18:30
shooter at exe one that has a couple
18:32
of levels that is ships I took
18:34
one of those levels and then imported a big giant landscape
18:37
mesh and put a bunch of foliage on it and I'm
18:40
gonna put down I gotta I gotta put down some
18:43
buildings and stuff like that with more pre-computed lighting So
18:45
it's a little more representative and
18:47
then I said hey calculate the light
18:49
mass inside this volume and Here's
18:52
the nav mesh volume and here's the physical You know, here's
18:54
the different the different areas that we need to care about
18:56
and hit build all to see what happened and it worked
19:00
And then I started looking at I started
19:02
changing the amount of foliage I started changing the amount of
19:05
pre-computed light to charge is for changing the amount of geo
19:07
in the world To see what it
19:09
did to the benchmark and got to a place that it
19:11
was it was interesting and working And
19:14
then I realized that I could change how long it
19:17
takes to run the benchmark By
19:19
changing the size of the
19:21
light mass volume the area that it cares about
19:23
getting good lighting for And
19:26
it's interesting because this process they like to
19:28
set up process isn't super threaded So it
19:30
favors high clock speed
19:32
low core count CPUs And
19:36
if you make the amount the area that
19:38
the light bake happens in small enough and
19:40
and not complex enough Then
19:42
that means you're gonna have a benchmark like I
19:45
could make this benchmark very easily favor say a
19:47
Quad core that runs at higher
19:50
clock speeds than say a thread ripper that's
19:52
gonna in practice be a much better Representation
19:55
of this benchmark and and I
19:57
started I started thinking about it was like everything that you're
19:59
creating content Like if you're making a premier benchmark, the
20:02
filters that you choose to run the transitions that you
20:04
choose to use all of that, whether it hits GP
20:06
or CPU, are inherently
20:09
political choices and it
20:11
determines like it
20:14
would be easy to make. And
20:16
I'm not suggesting anybody's doing this kind of, I don't think
20:18
anybody that's actually working in these spaces does do this, but
20:21
it requires a level of thoughtfulness that
20:23
I hadn't really considered before. Yeah. And
20:26
thoughtfulness into what can be a bit of a
20:28
black box. Like you see the same thing with
20:31
rendering out a video project, for example, right? Because
20:33
you're not only rendering with a video
20:36
codec or compressing with a video codec, you're
20:38
also compressing audio. You're also doing compositing of
20:40
different elements together and all of those things
20:42
are using different elements of the hardware and
20:44
different software routines and algorithms. And some of
20:46
those, like you said, same thing there. Some
20:49
of those things might be heavily threaded. The audio
20:51
and code might be single. In fact, most audio
20:53
encoders that I've seen are single threaded. But
20:55
same thing where you kind of
20:57
have to sit there and watch activity across different cores
21:00
as you're doing things to even get a sense of
21:02
like where the bottlenecks are and how to design around
21:04
that. Yeah. So what ended up
21:06
happening was I was loading up task manager and process
21:08
manager while the Unreal Engine, while
21:10
this was running and watching what was happening when.
21:13
And then I started thinking about, okay,
21:15
I need to like, because realistically, if
21:17
you're doing a production game, the setup
21:20
period is going to be a small fraction
21:22
of the actual light mass.
21:25
Like the light mass calculation stuff is
21:28
going to be the vast majority of your time
21:30
on this benchmark, even if you're using the GPU
21:34
light mass stuff, which
21:36
is another political choice because it's skewing in favor
21:38
of people who have big giant GPUs with such
21:41
and such. It's also saying that
21:43
anybody who has a GPU that has less than
21:45
I think 12 or 16 gigs of memory just
21:47
can't run the benchmark. Or
21:49
we have to use one of the third party
21:52
light mass calculations, which does support lower GPU RAM
21:54
configurations. So it's
21:57
a series of really... It's
22:00
an interesting series of choices. I don't know
22:03
that I have... I don't know that we're gonna
22:05
make this public. I don't know that I'd have to look at
22:07
what... I've agreed through license-wise to see if we even can. It's
22:09
a fair pain in the ass to set up and get running on it.
22:12
Okay. That's one of my questions, what's gonna be how portable
22:14
this is, because the nature of benchmarking in this setting is
22:17
you have to put it on a bunch of different machines,
22:19
right? Yeah. So that part isn't too
22:21
bad. I intentionally built
22:23
it without requiring any third-party UE
22:25
plugins, which is important. And
22:28
I use mostly stuff that I had
22:30
licensed to from... Let's
22:33
see, the UE store, the Quixel Bridge, stuff
22:35
like that. So it's like it should
22:37
be... I should... I think from a
22:39
content perspective, I'm allowed to distribute it.
22:41
I think Shooter on EXE is distributable
22:44
because most... Like I said, a lot of
22:46
people base their... That's the starting point for
22:48
a lot of first-person shooters, they use
22:51
Unreal. It's one of the reasons that
22:53
Unreal Engine games often feel like they
22:55
have same-y physics is because people start
22:57
with... Why reinvent? When
23:01
you start with a blank game, you're starting with
23:03
no physics and no gravity and no collision
23:06
and none of that stuff. And starting
23:08
with that stuff turned on is incredibly
23:10
helpful. Sure. I didn't realize they were
23:12
third-party... What's the term? Renderers engines for
23:14
calculating lightmaps
23:19
and stuff like that. I didn't realize
23:21
components like that were pluggable in the
23:23
Unreal Engine. I kind of assumed there
23:26
was one way to bake lightmaps and
23:28
it was their way. No,
23:32
there's a bazillion lightmap things. But there
23:34
are... I don't want to say infinite,
23:36
but there's a seemingly infinite
23:38
number of Unreal Engine plugins. Yeah, I
23:41
believe that. I've looked through the Unreal,
23:43
the asset store before, or as a
23:45
community that calls it the asset store.
23:47
Anyway, I know there's tons of stuff
23:49
out there for getting your game onto
23:51
Steam or Workshop, like mod
23:53
support, like plugins for... I guess the
23:56
distinction I'm trying to make is instead of adding
23:58
extra functionality that's not in replacing very
24:00
low level functionality. Oh yeah. That they ship with the
24:02
engine and that's what I was surprised that there are
24:05
options for doing. Yeah, so I
24:07
think that the main reason
24:10
the other Lightmass Calc exists, let's
24:13
see, it's Lu... it's
24:15
Lu Shong's GPU Lightmass Swarm
24:19
and it's a GPU Lightmass and
24:22
it's basically a
24:26
port of the GPU Lightmass accelerator that'll
24:28
run on 70 series cards because by
24:30
default, pretty much the Unreal Engine 1
24:32
only works on 3080s or 2080s or
24:34
3080s I think. And
24:38
it's because of, I believe, memory limits. I
24:41
can't remember. I see. So
24:43
that's the one example. Unreal treats
24:46
GPU Lightmass still as a beta thing. If
24:49
you're working in an office with a bunch of people, you
24:51
can run the Swarm controller. The CPU
24:54
Lightmass calculator will distribute across all
24:56
the computers in your network as
24:58
long as you have the Swarm controller installed on them and running,
25:01
which means it can go a little bit faster but it
25:03
still takes a long time. And
25:05
it means that you have to leave those computers running
25:08
if you want to take advantage of that. Sure. I
25:10
wonder how throughput limited that is. Not
25:12
very. It's not working those calculations. I assume the
25:14
amount of the actual amount of data that's going
25:16
right and forth is pretty small. It's just the
25:18
computation that is heavy. Yeah. So the thing that
25:20
I believe happens in the beginning is it breaks
25:22
up the levels into chunks and
25:25
then that's the thing that's not super threaded
25:28
or seems to be CPU bound or seems
25:31
to limit itself to a small number of
25:33
CPU threads, then
25:36
runs for a fairly long time, is breaking up the
25:38
level into discrete chunks and then it
25:40
distributes those out using that
25:42
Swarm client to the other machines on
25:44
the network. And they're not big. It's
25:46
just the math is complicated because it's
25:48
basically bouncing a bazillion rays
25:50
around the world. The difference
25:53
is we don't have to buy a big SGI workstation or some
25:55
Cray render or something to do it anymore. We can just run
25:57
it on a bunch of desktop PCs with 30
25:59
cores each. And are the days.
26:01
And the nice thing is you can set the
26:03
swarm thing so it just it's like to corps
26:06
which most people won't even notice when they are
26:08
losing their computers happier. Now
26:10
than whatever the reason to spend work is
26:12
interesting though is if you're was seen, environment,
26:14
artist or level designer or work in those
26:17
spaces new have to rebuild these levels frequently.
26:19
You'll often. Think. If you're doing
26:21
the good build of for the shipping levels
26:23
when the game is ready to move when
26:25
you're ready released to the gamer we are
26:27
updated new level in the game or whatever
26:29
you do the good lighting calculations right before
26:31
it ships right in that battle run for
26:33
hours and is decided to me. but if
26:35
you're making like. See. If you're changing
26:38
the position of a boxer, a door or
26:40
tree or something, you'll run a preview version
26:42
of Calyx more often than that because you
26:44
wanna see the shadow for the tree in
26:46
the right place. Mason it as an ear
26:48
for something else or or whatever and and
26:50
that can get fairly touches me. that kinda
26:53
takes you out of work. He would
26:55
give you have to do it during the day it is
26:57
a d d you do pre going for lunch. we do
26:59
it at the end of the day. It's because is going
27:01
to prevent you from doing anything for the rest of the
27:03
day now. so. Guy on
27:05
my door.kind of excited when you told me you're doing
27:07
this because some months ago with your sword on the
27:09
increases, this was not for this benchmark. It was for
27:11
shit, a game of the time. But you saw me.
27:14
As. Around the time of the line, computers remember
27:16
this oh yes well as baking lights You
27:18
just send me and didn't you send me
27:20
a screenshot of Task Manager. With. Every
27:23
single core on your thirty nine hundred K
27:25
pegged, flatlined, other than a hundred percent which
27:27
yeah, I couldn't even do And like. I've
27:29
been experimenting with different eighty one software encoders
27:31
recently. And. Even do a math stuff like
27:34
I haven't been able to get a to. I mean it'll
27:36
You know it'll in, load each toward pretty steadily. be still
27:38
see a bunch of spikes up and down. But.
27:40
But this, this was the only kind of
27:42
the only thing. Except maybe compiling
27:45
with the linux kernel or some some some
27:47
heavy sauces mothers and babies. maybe maybe some
27:49
meat eating as others have to do it
27:51
the like. I'd I have not
27:53
seen anything else besides that input of are
27:55
unreal calculations you're doing that. So.
27:57
Heavily thoroughly completely. pegs
28:00
each core all at once. Yeah,
28:03
so it seems like it seems ripe
28:05
for testing. And that was, and just
28:07
to be clear, that's even using the GPU light mass. It
28:09
still pegs all the cores. Like it
28:12
makes the, it made the power
28:14
cable coming into the computer get
28:16
warm. Wow. Just probably good, right?
28:18
Interesting. Yeah, so anyway, that's
28:20
what I've been working on. I'm curious what people think.
28:22
I'd be curious to your feedback. Like
28:25
I said, this is coming from a
28:27
place of wanting to mimic something
28:29
that's a semi-realistic. It's a little bit weird because
28:31
I think a lot of UE5 games are probably
28:34
not shipping with a lot of pre-baked lights.
28:36
They're probably doing global rendering, meaning
28:39
that the light is all the main light for
28:41
the world is dynamic. But,
28:44
you know, like we're doing fog and
28:47
other stuff that requires a little bit
28:49
of pre-compute. So it's, I
28:51
think we'll have something that's at least representative
28:54
of the performance you'll
28:56
get if not of the actual
28:59
game. Yeah, that's very cool. You know, at
29:01
first I was thinking to myself like, how big is
29:03
the audience for this kind of benchmark? It's limited to
29:05
the people who are working game developers, of course, but
29:07
you could see this, like, especially if it's something you
29:09
can release, you could see this just becoming
29:11
part of like some more standard benchmarking
29:13
suites. Like because it is a
29:15
legitimate real world, very heavy and
29:18
often very, very parallelized workload. I think
29:20
if I were releasing this for wider
29:22
than this one particular project we're working
29:25
on, I'd probably want to do like,
29:30
I think I'd probably want to do something
29:32
that's a little more representative of an actual
29:34
game. Like I think there's probably games out
29:36
there that would be willing to give us
29:38
access to like a level from the game
29:40
to do the calculation on. That
29:43
would be cool. Yeah, I guess to put it another
29:45
way, what I was trying to say, it's like, if
29:47
you look at a lot of the benchmarking suites that
29:49
are used in reviews for CDUs and graphics cards and
29:51
stuff, like they're Fairly robust, but
29:53
in some cases it feels like they're kind of
29:55
struggling to actually really load. modern hardware, you know,
29:58
and so having another. Real
30:00
World Something's not another son, just yet another
30:02
synthetic benchmarks but something that is a real
30:04
world workload the you can point out and
30:06
say here's how this actually worse absolute or
30:08
on is cool with I'm I'm excited that
30:10
the would more formula take ultimately video. It'll
30:13
be a video, will be part of
30:15
a video about a different thing. Occur
30:18
there and things are intentionally vague. I'm
30:20
sorry Marines. I look forward
30:22
to hear it. It's a thing about a saying
30:24
brad's yes or it shall We move on it
30:26
was hear about matter threat. I I'm sure does
30:28
get like I feel like it seems like I
30:31
haven't heard much syllabus. Every is going great I
30:33
bet swimmingly I am wrong with there's work. There's
30:35
only one. Might have
30:37
heard this one before. We need one standard
30:39
to unify all the standards it in. Today's
30:41
a month later Now there's only one Santa
30:44
right now. there's and with it's to number
30:46
seven. there were four plus wire. Something I
30:48
want. I don't think it's fair. This case
30:50
I seek. Okay, well or
30:52
had already has gone brand okay
30:54
might I drew a lot on
30:56
a i'm a kind of Urine
30:58
review in December that The Verge
31:00
ram. By Jennifer Pat
31:02
Patterson to eat or who apparently
31:05
has kind of been pushing matter
31:07
and thread to one and will
31:09
be taught thought the absurd explain
31:11
matter and thread again. We've covered
31:13
it. Fairly extensive yeah be bit
31:15
better. but the ideas that matter
31:17
and thread of threat is radio
31:19
sending matters a and interoperability kind
31:22
of communication standard said. it's theoretically
31:24
all of your devices, all your
31:26
smart home devices colloquial live cult
31:28
colloquially known as the Internet of
31:30
shit Ah, will. Toss each other and
31:32
worked with the with each other. Now. This.
31:35
Is a thing you could have done before if
31:37
you had some intentionality with your purchases via As
31:39
or even if he didn't Actually, especially if you
31:41
didn't that's how I ended up with homes her
31:44
son was. I've got a couple smart plugs for
31:46
my god because they advertise with us on a
31:48
podcast women I've got some others because we lot
31:50
of like a free pack on Amazon the was
31:52
on sale and and I've got yet a third
31:54
vendors plugs. as a christmas gift you
31:56
know what i mean we have some point that's how i came
31:58
home says and was just like And it's
32:00
like retroactively, I was like, I've got all
32:02
these competing things. I wish they all just
32:05
worked in one interface. So, yes, you could
32:07
and maybe will continue to do
32:10
this the old fashioned way
32:12
with something like Home Assistant rather than
32:14
standardizing it at the radio and
32:16
application level like Matter promised. I
32:19
mean, just to be clear, I have Matter
32:21
running on my Home Assistant box right now
32:23
and all the Matter and Thread devices in
32:25
the house, including the Apple TV and the
32:28
smart plugs that I have and the handful of other
32:30
things that I don't actually use except for Christmas time
32:33
because they're Christmas lights are
32:36
working off of the Home
32:38
Assistant Thread radio using the
32:40
Matter standard. Like, it's been pretty good for me. Yeah,
32:42
okay. That's fair. I've only used, like I
32:44
said, a handful of things and I don't like
32:47
the... So the interesting
32:49
thing about is it Thread? I always get
32:51
confused about which is which Matter. The interesting
32:53
thing about Matter is that they have the idea of this
32:55
edge router thing, right? The more
32:57
border routers. Yeah, where things that are
32:59
plugged in around your house like smart
33:02
speakers and digital picture frames and your
33:04
refrigerator maybe and made
33:06
by Google and Apple and Samsung all
33:08
can like expand the mesh of your
33:10
mesh network. Right. Like
33:13
the theory is that you would never have
33:15
to buy a dedicated controller or hub. Like
33:18
it's not like what I guess what SmartThings is or like
33:20
the Philips Hue Hub, stuff like that. You
33:22
know, like a thing that is just a little lump of plastic that serves
33:24
no other purpose than to unify all the actual
33:26
client devices. But that's also what Home Assistant is,
33:29
right? Oh yeah. Like that's what
33:31
I don't... I did not... When they initially pitched
33:33
this to people, that was not the thing I
33:35
took from this. The thing I took from this
33:37
was that all of the smart hubs would work
33:40
together on the same protocol so you wouldn't have
33:42
to have... Okay. So
33:45
I use Home Assistant to take all of the
33:48
different disparate stuff in my house and glue it
33:50
together into one application. And then
33:52
I'd dole it back out from there to
33:54
things like Google Home and Apple
33:56
Siri and Alexa if that
33:58
was a thing that I used. used so that
34:00
I could access them from the different places that I
34:02
wanted to do like voice assistance. My
34:05
understanding was that I shouldn't have to do
34:07
that step with Matter. That all
34:09
of the Matter devices should just be able to talk to
34:11
each other and like my home assistant would see HomeKit
34:14
and Google Play and all the
34:17
different Nest things and my
34:19
Nest things would see the HomeKit and the Google Play
34:21
and all the different standards. And
34:23
it seems like that's just not happening. That
34:25
is the promise. Every device is supposed
34:27
to work with every vendor's ecosystem and when
34:30
we say ecosystem we're talking Apple Home,
34:32
Google Home, Alexa, Samsung. What
34:34
is the smart things? Are
34:36
there other big ones out there? Yeah,
34:38
Philips Hue is one. Yeah, okay. I
34:42
don't think Lutron has signed
34:44
on but Lutron has a hub with their Cassetta
34:46
products. There's a
34:48
bunch of different places that this fires in. The
34:51
interesting, the good thing about
34:53
this was there were kind of two,
34:55
in reality there were a bunch of ecosystems, in reality there
34:57
were two, right? Because there was like HomeKit
34:59
which had some specific hardware requirements and
35:01
then there was everything else which was
35:03
more or less hubs using Zigbee or
35:06
some or Z-Wave or some variation of
35:08
one or the two. And
35:10
then Lutron had Cassetta which is on set of radios which is
35:12
quite nice but is off on its own. Yeah.
35:15
Anyway, this piece that Tui did was basically
35:17
kind of a year in review of spending
35:19
2023 testing Matter and
35:21
Thread devices and to be fair it sounds
35:23
like she's got so many things on her
35:25
network. It's like far, far beyond what any
35:27
average consumer would probably do. Well,
35:30
it's worth mentioning 2023 I think
35:32
is really the first year that Matter and
35:34
Thread stuff started shipping in any kind of
35:36
real quantity. Yeah, that's true. It's been almost
35:38
a year since I was looking at some
35:40
Nanoleaf Thread smart bulbs that had
35:42
I think were like kind of the first
35:44
Thread smart bulbs to the market. Yeah, the first
35:46
thing I bought was an Eve, what was
35:48
it I said it was, Eve Electric maybe? Yeah.
35:52
It was an Energy Smart Plug and I
35:54
bought it just to test the Matter implementation
35:56
on my home assistant box. Yes.
36:00
I think actually, let me see if I can find it
36:03
in the story. I think even, yes, it's the
36:05
Eve and Natalie for the two companies she
36:07
mentioned kind of bet
36:10
everything on Matter and Thread and are now pivoting
36:12
back to making non- making Wi-Fi
36:14
devices again, if that tells you how things are
36:16
going. Well, it's worth mentioning that the Eve
36:19
box that I got
36:21
started out as a Bluetooth device and
36:23
then got Thread updates over
36:25
time so that it was able to connect to the
36:27
Thread border routers. Yeah. Okay,
36:30
so it was a rough year.
36:32
Yes. Is the TLDR. Yes.
36:35
I guess I just kind of will address some of the main points raised
36:37
in the story and we'll link this. There's also a
36:39
couple of R's pieces that we read for this. Those will all be in
36:41
the show notes. But just kind of going
36:43
bullet point by bullet point of what Matter and Thread we're
36:45
supposed to deliver. Okay. So
36:48
like the first one is you can use any
36:50
company's device in any other company's ecosystem. That's the
36:52
idea. Like if you're an Apple
36:54
Home user primarily, but you end up with some
36:56
devices that are made for Alexa or whatever, theoretically
36:58
those are still supposed to work in the ecosystem
37:00
of your choice. And that works,
37:03
right? Sort
37:05
of, yes. It sounds like, I'm just going to
37:07
quote from the story here, it sounds like basically
37:09
at a baseline, yes, but
37:11
the actual functionality that you're going to get
37:13
is going to vary by which ecosystem in
37:15
and what features that ecosystem
37:17
supports. I mean, that's kind of how open standards
37:19
work, right? I guess so. Everybody supports
37:22
which parts of it they want and
37:24
everybody else supports the parts of it that they don't
37:26
support the parts that they don't want. So
37:28
like, is there a, for example here,
37:30
I guess? Yes, there is. I mean, you're not wrong,
37:33
but also as a consumer, I, that's the last thing
37:35
I want to worry about is well, do
37:37
I need to eyeball a table on the back of the thing to
37:40
know if all of its features are supported in home or what
37:42
Apple Home or whatever it is I like to use? Okay. I
37:44
mean, I will say this morning I got up and the power had
37:46
gone off last night and all of the light bulbs were on the
37:49
house and I had to open up
37:51
the Wyze app and go in and toggle on
37:53
each light bulb the, on power
37:56
off return to last state
37:58
rather than full blast. Because
38:01
at three o'clock in the morning when the power went off
38:03
and all the lights turned on in the bedrooms, it wasn't
38:05
great. Oh, good. Yeah. That sounds
38:08
fun. Okay. Yes. Here's
38:10
a quote talking about devices working in different ecosystems. If
38:12
you connect your Philips Hue smart bulbs to Apple Home
38:15
via Matter, you can't use Apple Home's adaptive lighting features.
38:18
If you pair any Eve Energy smart plug
38:20
with Amazon Alexa, you lose its energy monitoring
38:22
feature. If you add the GOVI's
38:24
Matter light strip to your smart home platform of choice,
38:26
you can't play with any of its cool lighting effects.
38:30
All of these things are able to
38:32
connect at a basic level and have
38:34
their basic functionality work, but higher
38:36
level stuff or anything that requires any vendor
38:38
specific support is not working across platform at
38:41
all. Okay. So I
38:43
can use adaptive lighting with Philips Hue
38:45
bulbs in my... I
38:48
have to do it from the Apple Home app. And
38:50
I can use... I have one of those Eve
38:52
Energy smart plugs. In
38:55
Home Assistant, I see the energy usage. In
38:57
the Eve app, I see the energy usage.
39:00
But if I ask Google about the energy usage, the Google
39:02
Home device isn't going to know what the hell I'm talking
39:04
about because it doesn't understand that. I
39:08
guess I don't see that as a problem. I
39:11
see that as... That's like saying,
39:13
hey, Google added a
39:16
bunch of proprietary shit to Chrome that
39:18
Safari and Firefox aren't going to support.
39:20
Well, but I mean, that analogy
39:22
would hold if there were some global browser alliance
39:24
that had promised all these browsers were going to
39:26
be able to interoperate. I mean, what's the W3C
39:28
then? Oh, geez.
39:31
That's a whole other episode. Yeah, yeah. Trust
39:34
me, I get up to my elbows in
39:36
enough half-based tech to absolutely understand why the
39:39
stuff wouldn't be working right at
39:41
a high level. But again, I'm coming at
39:43
this from the angle of... This thing
39:45
is supposed to take all the guesswork out of this process of
39:47
buying stuff. But the idea of this whole
39:49
standard is you're supposed to just be able to buy
39:51
devices and have them all work. And
39:54
the second I have to start doing homework to actually
39:56
figure out what that's going to mean in practice is
39:59
the second that... I'm just never going to
40:01
touch this standard and I think that's
40:03
probably where I'm going to end up at at the end of This
40:05
episode. I think I think we're in
40:07
the trough of despair, man Yeah, I
40:09
think I think this is like we've heard
40:12
the promise everybody got all excited about it started
40:14
coming out It's not living up to the promise.
40:16
Yeah I've got look I'll
40:18
cop to I'll cop to having been fairly excited about what
40:20
they were pitching for Matter and thread a year
40:22
and a half ago when we did that false
40:24
spot episode and it does not seem to be playing out
40:27
Okay, let's move on to the next point I mean is
40:29
this one is more pretty flatly damning I think matter
40:31
has something called multi admin mode, which I had not
40:33
been aware of but like it They
40:36
defined it in the spec enough to give it
40:38
a name Yeah That tells you anything and the
40:40
idea there is basically you can kind of the
40:42
inverse of what we're just talking about where it's
40:44
like You can use multiple products in any one
40:46
ecosystem of your choice in this case You can
40:49
use one product in multiple ecosystems So you could
40:51
have like I guess I guess the best
40:53
use case example is like You like
40:55
Apple stuff and you like to use the Apple
40:57
home interface, but your wife likes Alexa or Google
41:00
home ideas that the device
41:02
would show up in all of those Home
41:04
hubs on say everybody's phone and you could access
41:07
the basic functionality of the device from any of
41:09
those Sounds like that absolutely just does
41:11
not work. Yeah, I don't I'm
41:13
doesn't surprise me at all. I this one
41:15
this one again This
41:18
feels like not being intentional Like
41:20
I get that normal people aren't intentional
41:22
about computer stuff often But
41:25
all of this feels like hey you have to think about Think
41:28
about how this works and how
41:31
this is going to work Maybe this is just maybe
41:33
I'm just so damaged from years of covering this stuff
41:36
that I can't I can't Like
41:38
I yeah, I don't know. I'm kind of how I
41:40
feel. I'm kind of like don't cut them too much
41:43
slack I mean again I get I get in practice
41:45
why all of this stuff is such a huge mess
41:48
Also, my assumption is we're first
41:50
year like if I if
41:52
I have one argument with this One
41:55
complaint with the standard I'd say that they pitched something
41:57
that was way way way way
42:00
optimistic for something that's going
42:02
to have some really gnarly teething pains when
42:04
you're talking about five enormous worldwide
42:07
global companies that all want that
42:10
all agreed to the standard because they think they can own
42:12
it and like
42:16
yeah it doesn't surprise me that Phillips made a
42:18
thing that doesn't work with everybody yeah I
42:20
mean that that seems to be the case is that everybody kind
42:23
of wants to do their own thing even within a
42:25
spec that says they should all be working together so
42:27
I will say when I moved the
42:30
Eve thing from my home kit network into
42:32
the home assistant network which was a little
42:34
bit of a process I had to download
42:36
it I did I did go
42:38
to a setting on the app I had to
42:40
download generate a new pairing image
42:42
print that out take a picture of
42:44
my phone from the home Apple home
42:47
app it was a weird process yeah
42:49
but like I made the
42:51
choice to pair everything with home and
42:53
let it with a home assistant and
42:56
let it propagate out to home kit and
42:58
Google home from there rather than thinking I could
43:01
add it to the Apple home and have it get
43:04
sucked into home assistant and Google home from there like
43:06
oh yeah yeah I would never do that to be
43:08
clear as a home assistant user I also would add
43:10
everything through home assistant yeah I like I guess a
43:12
lot of my a lot of them a lot of
43:15
this the both the first two points are like I
43:18
don't think this is a problem if you I think
43:20
home assistant is the good implementation of all of these
43:22
technologies yeah I agree with you that I also still
43:24
think like home assistant has gotten way more accessible than
43:26
it used to be but I still don't think it's
43:28
within reach of the average consumer who doesn't want to
43:31
think about any of this stuff oh it's definitely nerd
43:33
shit yeah yeah yes extremely third
43:35
point is about the the thread border
43:37
routers that we were talking about where
43:39
basically your fridge or your light fixture
43:41
can be a device that routes traffic
43:44
from other clients instead of
43:46
a dedicated dedicated box sounds like that
43:48
is still a giant mess she says that most
43:50
of the devices on her network are still running
43:52
through dedicated boxes and that's specifically
43:55
I'll just quote this he
43:57
does are so thread border routers in my home formed
44:00
their own proprietary thread networks based on
44:02
their manufacturer and adults talk to each
44:04
other. Yeah, this is bad. Like, just
44:06
to be clear, this seems like either
44:09
ideally the apps should look and say, oh, hey, there's
44:11
three thread networks here. Do any of these belong to
44:13
you? Do you want to join one of the existing
44:15
networks? Any time you add something, so
44:17
you end up with one thread network instead of a
44:20
bazillion thread networks. There's
44:22
probably political reasons they don't want to do that because everybody
44:24
wants to own this network. The
44:27
benefit of having a big giant mesh network
44:29
with a bunch of border routers is that
44:31
you have everything connected to the same network
44:33
so that when you add something new, it
44:35
benefits from the breadth and width of
44:37
the net mesh and doesn't get
44:39
disconnected. This is both an onboarding
44:42
and installation and setup failure,
44:45
a communication failure. This
44:47
sucks all the way around. Yeah, for sure. Sounds
44:49
like a giant mess. Also, I haven't
44:52
mentioned so far, some manufacturers have kind of backed away
44:54
from this a little bit, like Belkin almost
44:57
a year ago, I think, announced that they were stepping
44:59
back from the whole thing and kind of leading. From
45:01
a matter of thread? Yeah, yeah, yes. I
45:03
mean, it doesn't surprise me. And other manufacturers
45:05
who are participating are kind of trying to, it
45:08
sounds like, carve out their own fiefdoms within
45:10
it. Well, but I mean, it's
45:13
quite literally saying, hey, we're going to take
45:15
the thing that makes this good and then
45:17
just not do it. Yeah, yeah, that seems
45:19
to be kind of how it's shaping up.
45:22
She talked about the thing you were talking about with the onboarding. She
45:25
says that the simple onboarding of you
45:28
can use any app you want and just basically
45:30
scan the matter code on the device and it's
45:32
just on, actually does generally work pretty well or
45:34
when it works, it's quite elegant. But I think
45:36
she gave the same example you did. You said
45:38
your thing fell off in that work and you
45:40
had to get it back on, right? Well, no,
45:42
no, I wanted to switch from the Apple Hub
45:44
to the Home Assistant Hub. It was switching, I
45:46
swear. Yeah, because just to be clear, sucking
45:49
it from the Apple Hub into the
45:51
Home Assistant Hub wasn't working particularly well,
45:53
which meant that it wasn't accessible in
45:55
things like Google or Alexa. So,
45:57
yeah. But
46:01
I think my assumption, and this
46:03
may be naive, but my assumption
46:05
was that at least for this
46:07
first generation, maybe two generations, probably
46:10
you're going to have to use some proprietary app
46:12
to do most of the configuration stuff for everything.
46:14
That's kind of exactly where she's at here. She
46:17
gives an example about some TP-Link TAPO light
46:19
switches that she could not get
46:21
to connect to the network
46:24
for Apple Home because they're TP-Link. She
46:26
says, I quickly defaulted to the TAPO app where they
46:28
paired immediately. When I went back to add them
46:30
to Apple Home and then Google Home through Matter, two
46:33
of them dropped offline on both platforms but continued to
46:35
work fine in the TAPO app. So
46:37
yes, it's still a lot of vendor
46:39
siloing. So
46:41
I have some of those TAPO switches that I use for
46:43
like Christmas tree and Christmas lights around the house. And
46:46
I had the same exact experience until I paired
46:49
them with the Home Assistant Matter
46:51
network and then it was fine. Your
46:54
mileage may vary, I guess. Yeah. And
46:57
then she talks about, you mentioned some of your
46:59
devices have upgraded to Matter that used to be
47:01
like what, Bluetooth? They were Bluetooth and ZigBee or
47:03
Bluetooth at HomeKit, I think is actually what they
47:06
were, which is really good ZigBee pork, I think.
47:08
I can't remember how Apple does that. She
47:11
says, for example, her Philips Hue Hub has gotten
47:13
less reliable since she did the upgrade to Matter
47:15
support for it. Like,
47:18
yeah, this is
47:20
the thing. The problem with this kind of thing
47:22
is everybody is using, it's up
47:24
to everybody's implementations, right? And
47:27
if your hub has a bad implementation, if your
47:29
endpoint has a bad implementation, if
47:31
you're like, yeah, I just,
47:34
I'm bummed. Yeah, it is a bummer. I
47:37
was hopeful about this stuff. Although again, because we're
47:39
deep into the Thomas isn't rabbit hole, it kind
47:41
of covers that use case. Well,
47:43
it covers that use case but it also, like
47:46
I have ZigBee Z-Wave and HomeKit. No,
47:48
I don't have HomeKit. I have ZigBee
47:50
Z-Wave and Matter and Thread all
47:52
running on the same box. It's in the center of my
47:54
house. It's powered with power
47:56
over Ethernet. It's relatively easy to access and
47:58
maintain. But mostly I just let it
48:01
sit there and I update it every couple of months when it
48:03
gets an update and then I don't think about it too much.
48:06
And in that regard, it works exactly the way I
48:08
want for this kind of stuff I just don't want
48:10
to think about. I just want it to work. Yeah.
48:12
I mean, the one aspect of this that is not
48:16
supported by the home assistant use case is the
48:19
onboarding, the initial setup for a new device.
48:21
Yeah. Because you're still even with home
48:23
assistant, you're still, well, unless it's matter, it sounds like if
48:26
it's matter, you're going to add it straight into home assistant,
48:28
right? Like you mentioned. It has to
48:30
be matter out of the box. But
48:32
still, I think I would use the
48:34
on the official app to set up
48:37
the pairing would be my guess. That's
48:39
the thing I cannot stand. Honestly, that's the thing. That's
48:41
the aspect of all of this that
48:43
annoys me the most is just having like half a
48:45
dozen different smart vendor apps on my phone and having
48:47
to get a new one every time I buy somebody
48:49
else's product. Oh, so I just put them on a
48:51
folder and then never think about it again. Yeah. I
48:54
mean, if you don't have to make an account in one, it's way better. But some
48:56
of them... Well, okay. I
48:59
don't really buy anything that's cloud mediated, so I'd never have
49:01
to do an account. Yeah. I mean,
49:03
well, generally, once you do set up, once you might
49:05
want to check and make sure it... I'm
49:08
going to look at Wwise. I think Wwise might have an account.
49:10
I can't remember, but those balls were pretty good. They were really
49:12
cheap. I think I'm pretty sure TP-Link Casa has an
49:14
account. I think the Belkin WeMo stuff
49:16
does as well. It's not it's not
49:18
routing like control signals for the devices
49:20
through the cloud. But it's... Yeah.
49:24
But you do have to make an account just to get past
49:26
the app onboarding to let it actually pair something. I have a
49:28
Casa account. I have a Wwise account. Yep. All
49:30
that stuff sucks. It doesn't bother me. I try not to worry
49:33
about it too much. So
49:35
the thing that I found, I was
49:37
early on Zigbee and bought a whole
49:39
bunch of Cree Zigbee bulbs, which were
49:41
great. When they started clunking
49:43
out about 10 to 12 years in, I
49:47
replaced them all with Wi-Fi bulbs because
49:49
the Wi-Fi... It turns out, I
49:51
spent a fair amount of money and time building a
49:53
really robust Wi-Fi network that covers the entirety of my
49:55
house. And it
49:58
was infinitely more reliable than Zigbee. radios
50:01
in what are often
50:03
met metallic fixtures that
50:06
would block Zigbee with radio signals. And
50:09
as a result, I would have bad
50:11
radio drop off on the far side of the house
50:13
and like light bulbs would just stop working. So once
50:17
I switched to Wi-Fi, I haven't had that problem. Yeah,
50:19
that's fair. I mean, I guess your average house is
50:21
just inherently going to have way better Wi-Fi coverage. Then
50:23
it's not like you're going to put multiple Zigbee radios
50:25
all over the house, right? Although, or it is a
50:27
mesh network. So I guess... I mean, theoretically, each of
50:29
those bulbs should have been a note on the mesh.
50:31
In reality, there's
50:33
a limited number of hops you can do with
50:36
Zigbee. Zigbee is, I think, better than Z-Wave. Z-Wave
50:38
is like four hops or something ridiculous. But
50:41
in practical terms, when
50:44
somebody would walk across or sit in a different place
50:46
in the house, the light at the
50:48
far corner of my daughter's bedroom, which was as
50:50
far from the Zigbee radios you would get, would
50:52
just stop responding. And it happened
50:55
often enough. And then because of
50:57
the age of those bulbs probably, when it
50:59
will stop responding for a lost signal in
51:01
a frequent enough period of
51:05
time, it would eventually need a reset to come
51:07
back on. And I got tired of
51:09
doing the leave it on for a second,
51:11
leave it off for two seconds, leave it on for a second, leave
51:13
it off for two seconds, leave it on for a second, leave it
51:15
off for two seconds, thing five times
51:17
to reset the bulb. So
51:20
I just replaced them all with Wwise bulbs. Yeah, that makes
51:22
sense. All of that is good to hear. I did
51:25
talk to some people on the Discord and the Home
51:27
Automation channel, and there are definitely some fans of Zigbee
51:29
in there. I think Zigbee is generally
51:31
pretty good. I don't know, man. Maybe
51:34
I should just stick with Wi-Fi stuff at this point.
51:36
Here's the thing is, if I had a bunch, if
51:38
you have, if you're in a
51:40
situation where you can put Zigbee switches in
51:42
your wall, but serve as mesh nodes, as
51:45
mesh routers, I think it's pretty
51:47
solid. The wiring in my
51:50
house is old. So all of our wires are
51:52
two pole wires. This is not a neutral wire.
51:54
So that means I couldn't run
51:56
any modern switches basically in this
51:58
house. So, there's
52:01
a Luchron Cassetta makes some
52:03
two wire switches. They
52:06
are not Zigbee. There were
52:08
some GE Z-Wave switches that were
52:11
two pole wires that were a little
52:13
scary. They don't make those
52:15
anymore. They're really hard to find. They got really expensive
52:17
after they stopped making them. That's
52:20
pretty much it. So, the option
52:22
for Zigbee mesh nodes was big
52:24
giant wall warts that dangle out of the
52:27
wall to plug appliances or other
52:29
things in. And I ended up just
52:31
putting a couple of those in like disused plugs in
52:33
corners of the house so that
52:35
I would get a solid enough network so that
52:37
my daughter's light would turn off at bedtime. Literally
52:39
just using them as repeaters. You don't even have
52:41
anything in them? Straight up just repeaters,
52:43
yeah. So, I was just burning
52:45
electricity to make the radio better. And
52:48
like I said, the Wi-Fi network's real good. So... Yeah.
52:51
Do you have any concern about... I don't know how...
52:53
Well, you can speak to this. If you reach a
52:56
certain critical mass of internet of shit, devices on your
52:58
Wi-Fi, like at some point your Wi-Fi network is just...
53:00
The Wi-Fi radio, I should say, is fairly crowded.
53:03
Yeah. The bulbs are low traffic enough that I
53:05
don't really care. That's fair. Like I can open
53:08
up my UniFi console and tell you how much
53:10
traffic the average bulb
53:13
does. But it wasn't... When
53:15
I looked, it was not very much. It's probably not
53:17
a lot. I mean, there's got to be some kind
53:20
of like fairly regular keep alive kind of signal
53:22
going back and forth though, right? I think it's
53:24
like... I think it's like... I don't
53:27
think it was very many packets. Hold on. Probably
53:29
not. Maybe I'm overthinking that aspect of it. Or
53:31
maybe... I mean, that was a
53:33
concern that was raised on the Discord, was not cluttering
53:35
up your Wi-Fi networks with this stuff. So,
53:37
one problem I will tell you is I'm
53:39
getting close to the number of... I'm going
53:42
to have to move into a second subnet
53:44
before too, too long. Interesting. Wow. I
53:47
mean... Because I'm going to be more than 200... Well... What's
53:50
the size of your address range? I
53:52
mean, there's... Talking of what? 8-bit subnet
53:54
mask? Yes. What is that?
53:57
255 addresses? Yeah, but I have some... I
54:00
fixed IPs reserved at the front for reasons
54:02
that I don't remember. I
54:04
don't remember what they are now. So I'm afraid
54:06
to mess with them. Maybe you can just release
54:08
them and see what happens. Nothing breaks. You're fine.
54:10
Also have, uh, I
54:12
also have, okay. So here's a wise bulb. I can
54:14
tell you how much data it uses now. Um, it
54:18
has in the last 16 hours
54:22
and one minute, it's done. I'm trying
54:26
to find, oh, uh, a
54:28
hundred and 1250 packets for 101 kilobytes down and 4,490 packets for
54:31
270 kilobytes
54:36
up. That's not a lot of data. I
54:38
mean, I wonder, I wonder if it's more of the
54:40
interval that it's actually touching the radio rather than the
54:42
amount of data that it's sending that matters there, but
54:44
either way that doesn't seem like much. Yeah, it's, it's,
54:46
it's not much. The bigger problem is the congestion on
54:48
the, on the, um, on the, um, on
54:51
the subnet. So yeah, you can put it, you
54:53
can put all this stuff in a separate VLAN
54:55
if you wanted. I mean, I, again, like we
54:57
mentioned at the top, he probably should do that
54:59
anyway. I've been meaning to do that for ages
55:02
and just haven't done the homework. So I generally
55:04
just buy stuff from reputable places. I don't buy
55:06
a bunch of like sketchy Teemu
55:08
or Ali express wifi devices,
55:11
things devices. Um, but
55:14
yeah, I mean, there, there is a lot
55:16
of this stuff has firmware that's released and
55:18
then is never looked at
55:20
again and security vulnerabilities aren't patched or
55:22
whatever. So it is a, it
55:25
is a concern. Yes. Um, everything I've got
55:27
is it's all I devices, Belkin and TV
55:29
link, except for I've got one kind of
55:31
no name. Uh, it
55:33
is a home. Well, it's home kid enabled. It
55:36
isn't I Haper. I think it's
55:38
how I, I already shipped that it's probably HAPER, which
55:41
is now defunct. Like you can't even find anything about
55:43
that company. I've had it for like five years. It's
55:45
like kind of a, a no name light strip led light
55:47
strip, but it is home kid enabled. Unless
55:50
they found some way to spoof the home kid
55:52
or required home kid hardware. I
55:54
assume it is legit enough that it's probably fine.
55:56
Um, but you're, yeah, you're right.
55:58
I think most of the stuff I have. is pretty reputable as
56:01
well. I have thought about
56:03
I have thought about replacing that light strip with the, you
56:05
know, you can just buy the kind of light strip
56:07
by the foot LED strips by the foot. I love
56:10
those and cut those and get like the little ESP
56:12
32 controllers. Uh huh. Kind of I
56:14
thought I have thought about getting away into that stuff.
56:16
But at any rate, we just had
56:18
an earthquake. Whoa, I didn't feel
56:20
anything. Live earthquake reacts. What
56:23
is your what is your go to? What's your
56:25
go to geological source? Usually go to earthquake. I
56:28
go to USGS. First I post on blue sky
56:30
was there just an earthquake. That
56:32
was my method for years was run to Twitter and see
56:34
if everybody's going to earthquake. Like I
56:36
want my last thing, my last
56:39
thing that happened to be me asking if there
56:41
was an earthquake. Um, I
56:43
just, it sounded like a bus driving by and there was
56:45
a little bit of a rumble, but
56:48
I didn't feel a whole lot. I
56:50
could have gone. It's been a dust bus driving
56:53
by you use GS. Yes.
56:55
Where is San Juan Batista? That's pretty far
56:57
south. It's like by happening Bay. Uh, closer
57:00
to me than you though. Oh yeah, it is. Yeah,
57:02
that is down past Santa Cruz. 7 14 UTC. Oh,
57:07
I didn't, I didn't look at the time.
57:09
Oh, it's 10 minutes ago. No, that's not
57:11
it. Okay. Well, who knows? We're
57:14
still in the window for it to be on though.
57:16
It wouldn't, it wouldn't have hit. Yeah. It usually takes
57:18
a minute or two. Stay tuned as earthquake watch continue.
57:22
Um, what were we talking about? Uh, those
57:25
little light strips, the little, there's, there's a
57:27
whole scene around like open source firmware for
57:29
those little ESP controllers. There's so much fun
57:31
stuff you can, like if I next, next,
57:33
when I move into the next house, uh,
57:36
I'm going to run those all
57:38
under the edge of the kitchen cabinets. Yes. Cause
57:40
I want more lighting on my old and my
57:42
eyes are bad and I want lighting there. Great.
57:44
Yeah. To be able to, especially if you set
57:47
it up on like motion sensors or even if
57:49
it was just on your phone or standard stuff,
57:51
but like having like nice out of the way
57:53
under lighting where you're not actually staring at the
57:56
bulb, but you still get the good kind of ambient
57:58
bounce light and off under. And
58:00
stuff have you seen have you seen the new hotness
58:02
on the internet of shit stuff? No
58:05
presence sensors that use Wi-Fi
58:07
radios signals as radar
58:11
So they can tell like where you
58:13
are in the house where you are in the room
58:15
Oh, wow, whether you're like laying on the sofa whether
58:17
you're sitting up whether you're awake or asleep. Oh It's
58:21
a little creepy Definitely
58:23
put that on its own VLAN. Well,
58:25
yeah, cuz it just needs ambient Wi-Fi It
58:27
doesn't it doesn't need it's not actually using
58:30
data. It's just using the radar to radio
58:33
as radar It's basically
58:35
replacing motion sensors. There's a little like puck things
58:37
you put in the ceiling. That's kind of cool
58:40
It's honestly it's the thing I've always
58:42
wanted because they're more sensitive and they're like
58:44
sensitive enough to detect moving a mouse or
58:47
something Because often I'll be in my
58:49
office and the motion sensor isn't sensitive enough to detect
58:51
that so it just turns off the lights in Here
58:53
after a while and I would I'm thinking I'm gonna
58:55
put some in the in the in the house I've
58:59
never really been big on setting up automations against motion
59:01
sensing because I just feel like there are way too
59:03
many edge cases where I Wouldn't want the thing. It's
59:05
not great Like every single time you open
59:08
a door having a light come on like I don't think I want
59:10
that actually Well, so I have I
59:12
have a door sensor on our back door every time you
59:15
go out the back door and it's after dark The back
59:17
porch light turns on which is awesome Physical
59:19
switch right next to there. So if I don't
59:22
want that happen, you just flip the switch off.
59:24
Okay, you go out Yeah, that's good. The other
59:26
the other one was when the kiddo was young
59:29
having Having a
59:31
motion sensor and are the master bedroom where we take her out
59:33
of the bath To dry her off and
59:35
get her dressed for bed Was incredibly useful
59:37
because you have a baby and a bunch of shit
59:40
in your hands And you just walk in and the
59:42
light turns on that was that was really nice. Yeah,
59:44
there are these cases But that's what I got. Yeah
59:47
home automation. It's yeah still
59:50
In progress. I so here's the
59:52
thing We're at
59:54
the end of basically the first year or
59:56
18 months that the stuff is shipping. Yeah.
59:58
Yeah, I should I should say the right
1:00:00
up from 2B on the Verge and also
1:00:02
the Kevin Purdy on Ars Technica. I read
1:00:04
some stuff from him. Like both of them
1:00:07
conclude by saying, yes, this stuff is
1:00:09
all a giant mess, but like, we're still hopeful that this
1:00:11
is going to be the standard of the
1:00:13
future in X number of
1:00:15
years. The thing I will
1:00:17
say is it took a good 10 years before
1:00:21
Z-Wave was usable at all. Like
1:00:24
it was, it launched, it was terrible when it was
1:00:26
new. ZigBee took a similar
1:00:28
length of time. These
1:00:31
are, the interoperability adds
1:00:33
a level of complexity that goes far
1:00:35
beyond what ZigBee and Z-Wave were able
1:00:37
to do. I think like
1:00:40
the initial bug, that there are bugs
1:00:42
with complicated things like the
1:00:44
mesh, the devices
1:00:47
connected to the mesh spreading out from
1:00:49
one vendor to another vendor's networks
1:00:51
and vice versa. It doesn't surprise me.
1:00:54
I think it's really, really unfortunate that
1:00:57
the hubs, that the default for every
1:00:59
single edge
1:01:02
router isn't, hey,
1:01:06
is there a thread network here that I need to connect to?
1:01:09
Maybe there is, and then connect to that rather
1:01:11
than just setting up their own stake in the
1:01:14
ground. Yeah, that's terrible. Like that
1:01:16
is a dopey decision that should be rectified.
1:01:18
Yes. I will say though,
1:01:20
the danger of all these, even if these
1:01:22
problems are understandable at this early date, the
1:01:24
danger is creating this kind of retroactive chicken
1:01:26
and egg situation where because it's a mess
1:01:28
early on, nobody adopts it, nobody buys in,
1:01:30
and the manufacturers get scared and they all
1:01:32
back off from it. Not worried about that
1:01:34
at all. No. Nope.
1:01:37
They've sold enough stuff already, we're fine. People are going
1:01:40
to be, yeah, because here's the thing. It's
1:01:43
in the manufacturer's interest for this to
1:01:45
work. Because then, because if
1:01:47
it doesn't, if
1:01:50
there's competition and there's room for other
1:01:52
devices on your networks, then
1:01:54
there's opportunity for Apple to eat
1:01:56
and to Google and Samsung Share
1:01:58
and Google and... Samsung to eat into
1:02:01
Apple's share. If this doesn't
1:02:03
work, then that opportunity goes away and
1:02:05
there's just gonna be two silos forever
1:02:07
and ever. And there's never an
1:02:09
opportunity for Google to capture iPhone users
1:02:12
and Apple to capture Google users and
1:02:14
the silos get deeper and deeper. So
1:02:17
they're gonna continue pushing it and they're right to.
1:02:19
It's gonna take a while. It's gonna suck for
1:02:21
a little bit. It'll be fine eventually. Read reviews.
1:02:23
Brenton Boucher, Jr.: Yeah. So it just leaves you
1:02:25
in a weird limbo right now if you're looking
1:02:28
for new stuff. I'm not sure what to get. I
1:02:31
will say is a much smaller
1:02:33
sample size than the people that we talked
1:02:35
about who've read about this have. But with
1:02:37
the handful of devices I've used, the home assistant
1:02:39
implementation has been pretty good and
1:02:41
it just works with the other
1:02:43
stuff automatically. So your mileage may vary,
1:02:46
I guess. Brenton Boucher, Jr.: That's promising.
1:02:49
that's as good a place as any, I think, to wrap it up
1:02:51
this week. Please let us know what you think about the twofer. If
1:02:55
you do, we will do these more often. If you like these, we'll
1:02:57
do it more often. I think it's nice to be able to talk
1:02:59
about things that maybe couldn't sustain a
1:03:01
whole episode on their own. Brenton Boucher, Jr.: Yeah, I
1:03:03
agree. be back next week with
1:03:06
more of Bren Will Made
1:03:08
a TechPod. As always, this podcast is
1:03:10
brought to you by the listeners. You. Brenton Boucher,
1:03:12
Jr.: It's true. Thank you,
1:03:14
listeners. Brenton Boucher, Jr.: You did this. is
1:03:17
your fault. We wouldn't be here if it weren't for you. And it's
1:03:19
especially true of our patrons. So
1:03:21
we don't take ads. We don't take ads. The
1:03:23
only way we make money on the show
1:03:25
is by the generous contributions of almost
1:03:28
3,000 patrons. So thank you
1:03:31
so much, patrons. And as
1:03:33
always, if you'd like to find out how
1:03:35
to support the show, you can go to
1:03:37
patreon.com/tech pod, where for five bucks a month,
1:03:39
you get access to the fabulous TechPod discord,
1:03:41
which is full of beautiful nerds, even though
1:03:44
they're advocating the use of Z-Wave, which I
1:03:46
strongly disagree with. To be fair, I
1:03:48
think it was like one person who was in on
1:03:50
Z-Wave. Most everybody was into Zigbee.
1:03:53
I'm not going to pick on the Zigbee or the Z-Wave people, but I'm
1:03:55
just going to say, like we
1:03:57
sometimes disagree, but I still love each and
1:03:59
every person. there. And
1:04:02
for, like I said, five bucks a month, you get
1:04:04
access to that. You get the patron exclusive episode where
1:04:06
we've been threatening to talk about hell divers, but haven't
1:04:08
yet. And and oh,
1:04:10
my God, the fucking this flying enemy thing
1:04:12
is so amazing. I haven't seen the flying
1:04:14
enemies yet. I haven't seen the enemies either.
1:04:16
But the way they are handling it is
1:04:18
you mean the the the fact that we're
1:04:21
forcing the bugs to evolve by killing them
1:04:23
more efficiently with the gas? Specifically, it's this
1:04:25
weird, plausible deniability stealth update thing they're doing,
1:04:27
or they snuck them into the game without
1:04:29
telling anybody. And then when people started finding
1:04:31
them and it seems like they're finding them
1:04:33
sporadically, like the spawn rate is quite low.
1:04:36
And then the developers are basically denying that they're in the
1:04:38
game. I don't know how
1:04:40
I feel about the developers gaslighting players. I
1:04:42
think it's I mean, considering it fits
1:04:44
with the brand, it fits with the game.
1:04:46
Considering the political milieu that this game is
1:04:48
lampooning, I think I
1:04:51
think propaganda that turns truth on
1:04:53
its head probably makes perfect sense.
1:04:55
Did you see the propaganda on
1:04:57
YouTube where somebody who's data miners
1:05:00
pulled an upcoming video that explained
1:05:02
spoilers, I guess, for Helldivers two, if you care
1:05:04
to come back in 45 seconds. The
1:05:07
bugs, the gassing that we're
1:05:09
doing of the bugs right now is causing
1:05:12
them to evolve. So we're going
1:05:14
to get new varieties of bugs that are
1:05:16
super bugs because of the gassing that's happening
1:05:18
across all the planets. It
1:05:20
is backfired. They seem like they
1:05:23
went into this game with some ideas. They
1:05:25
said this game is most
1:05:27
ideas per per gigabyte.
1:05:30
I've seen in a video game in a
1:05:32
long time. Yeah, it's pretty exciting. Anyway,
1:05:35
thanks to all of our patrons. You can
1:05:37
go to patreon.com/check pod. Like I
1:05:39
said, five bucks a month. We appreciate everyone's support.
1:05:41
But we especially appreciate our executive producers, your patrons,
1:05:44
including Paddle Creek Games, make us fracture, which
1:05:46
is out now. You should check it out.
1:05:49
I mean, it's pretty good. Oh, cool. Andrew
1:05:51
Slosky, Jordan Lippitt, tech buddy crimes
1:05:53
dot com. That's a URL now.
1:05:57
Just wedge Joel Krauska, Twinkle Twinkie David
1:05:59
Allen. James Kamek, and Pantheon,
1:06:01
makers of the HS3 high-speed 3D
1:06:04
printer. I'm sorry, can you repeat that
1:06:06
URL again? Tech.buddycrimes.com. Okay.
1:06:09
That's pretty good. That's pretty good. Did you
1:06:11
go to it? Maybe. I'm
1:06:13
going now. That'll do it for us this
1:06:15
week. Thanks, everybody, for your support, and
1:06:17
we will see you all next week.
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More