bvisness:
| https://handmade.network/forums/jobs-board/t/7186-simple_computer_company#22150
| Handmade.Network
| Simple Computer Company | Jobs board | handmade.network Forums
| Computers should be simplified (whole stack)There aren't good laptops on the marketNo company see…

oliver:
thats pretty cool

dotbmp:
I think "simple computer" is a pretty marketable concept

bvisness:
yeah, I think Chromebooks are a good example of that

dotbmp:
to consumers it looks like chromebooks
to devs it looks like heaven
| https://www.sifive.com/blog/incredibly-scalable-high-performance-risc-v-core-ip
| https://www.sifive.com/share.png
| Incredibly Scalable High-Performance RISC-V Core IP - SiFive
| Introducing the new SiFive U8-Series Core IP
hold on how the fuck did they get 7nm manufacturing...?
or is this just IP

ratchetfreak:
or it's a smaller die

dotbmp:
anyway to sort of continue the conversation from mollyrocket the other day, chromebooks are a good example of not really needing backcompat for new platforms
granted chromebooks have the web and thus all of web shit
but browsers are a relatively small problem
just plop webkit into your platform
or
cringe
chromium

bvisness:
curious what OS this would be
Chromebook gets off easy because everybody knows how to use a web browser

dotbmp:
not an existing OS

bvisness:
good luck running software though

dotbmp:
indeed

bvisness:
people are gonna want it to be useful

dotbmp:
my take is that new OS right off the bat is not the correct approach
new platform, sure
as a portable layer on top of existing OSes
and then once you have a market, you build down to the hardware

ratchetfreak:
but picking an existing OS as the kernel API ties you to that OS

dotbmp:
not exactly
you can abstract over it
of course you have to be careful not to let existing OS design decisions influence your own to the greatest degree possible

ratchetfreak:
that's not how public facing apis work

bvisness:
I almost feel like you'd have better luck virtualizing the whole thing, but that is not exactly a very Handmade OS approach

dotbmp:
I mean yeah you could always do an OS on top of a VM
| https://urbit.org/
| Urbit - Your personal server
| Urbit is a simple, durable system for running a personal server that you own 100% and forever.
like this
but yeah I'm not super thrilled about that idea

bvisness:
that urbit thing is some serious galaxy brain action
I do actually think there is room for some real innovation in what personal computing looks like given the broad availability of the internet, but I don't ever really want to demote computers to dumb terminals
I want to benefit from strong connectivity when it's there, and still have a robust computing experience when it's not

dotbmp:
I think "home servers" would be a sweet spot
Given the tiny computing demands of most households

bvisness:
that is definitely a thing I'm interested in
like, here's a raspberry pi in a slick case that you can plug into your router

dotbmp:
It would be less expensive to buy a server box you plop in a closet, than two desktops, four laptops, five tablets, whatever
Instead, those things could be thin clients
You could upgrade it as needed
Like, why have a game console... just buy a "console upgrade kit" or whatever and install it to your server
And play from any connected display in your house

AsafGartner:
Home server with physical cartridges per app, so they can bring their own memory/storage/co-processors.
So they don't have to compete with other apps that you have installed.

dotbmp:
Bwahaha

bvisness:
I don't think the in-home streaming situation is actually good enough in general to support something like that
maybe it's just implementation, but I couldn't get my Steam Link to stream well over wifi in the same room as my computer and router
until I went to 5GHz that is

AsafGartner:
Imagine if Google Photos was a cartridge with an SSD and an AI core for automatic tagging.

bvisness:
maybe just interference in my building, but still

dotbmp:
Put some USB 4c cables through the walls :^)

AsafGartner:
No more accidentally letting other people download your videos.

dotbmp:
Imagine if no Google Photos

AsafGartner:
But how would you share baby photos?

dotbmp:
If you have an always-on home server you can just tunnel home on your phone

bvisness:
I think online photo features are really useful and I wish companies weren't scraping them all for data!

dotbmp:
Sharing is more of a concern that does involve centralized services
(in the absence of a revolutionary distributed system)
But it could be much more selective than "online by default"
After all people don't want to share most of their photos
Well
Most photos are taken to be shared, but as far as photos taken "offline" rather than as part of an app
And most are not final takes, either
If you take 30 photos at a family event you probably want to share 4-5

bvisness:
A good photo app is one of the apps I want to make someday

dotbmp:
And nudes are becoming a massive problem in terms of cloud services
Imagine opening your Google Photos and having someone's amateur sex tape on your phone hahaha
What a farce

bvisness:
all the photo services are very proprietary and closed-off and I rue the day when everybody loses all their family photos because Google shut something down

dotbmp:
The trust is the bigger problem imo
If people were actually nervous about these services failing them it would be less bad
But we take this stuff for granted

bvisness:
I want the benefits of my photos being available on all my devices, but I also want to be able to just find a jpg file and email it to somebody
things like the Apple photo library (on Mac) are so egregious about this
you have to find the export button or whatever

ctp:
i would bet theres a market for a gitlab-like 'host your own cloud service' type thing for personal email+photos+docs+sharing

AsafGartner:
| https://owncloud.org/
| ownCloud
| ✔ ownCloud - The leading OpenSource Cloud Collaboration Platform.
| Access & share your files, calendars, contacts, mail & more from any device; on your terms.

bvisness:
it has to be easy to set up though, and that's where I think it usually breaks down

dotbmp:
Well let's take the home server thing
Let's say it becomes popular
Or even marginally popular
Maybe there are 30K people with them in the US
They're always-on
You keep your photos on them on redundant storage with backups
You could have opt-in peer-to-peer backup "coalitions"
Just a bunch of people who agree to share some space on their storage for and some CPU time/bandwidth to do distributed redundancy through peer-to-peer streaming

bvisness:
some home backup programs have done that but they are inevitably frustrating (not because it's inherently frustrating, but because it's made poorly)

dotbmp:
Right, this would have to be a first-class feature
(When I say home server I mean with a new software platform etc)

ctp:
shouldnt depend on peer-to-peer backups imo. "can anyone seed [picutres of my kids - siterip] im at 99%

AsafGartner:
I wouldn't mind a regular backup service. You can encrypt your files locally on backblaze for example.

dotbmp:
@ctp this is a reductive view
I'm not talking about using bittorrent
you'd opt in to a group of thousands of people doing this

ctp:
personally I would never depend on a peer to peer system for backups. your backup can be encrypted and stored on normal remote services.

bvisness:
I don't think I share the same concerns but I also think that I generally wouldn't care about peer-to-peer

dotbmp:
I literally said
You keep your photos on them on redundant storage with backups

ctp:
has any peer to peer system reached the level of reliability required to use as a backup? not talking about cryptocurrencies

dotbmp:
these aren't mutually exclusive concepts

bvisness:
I don't think remote centralized stuff has to go away, especially for the backup case

dotbmp:
I disagree
kind of
it's fine most of the time

ctp:
backups are a reliability problem, not a privacy problem or a centraliztion problem.

dotbmp:
but when it's bad, it's really bad
(see: the Fappening)
a combination of local and remote backups is a good balance
if your drive fails, you have a backup locally
if your house burns down and your server is destroyed, you have a remote backup

ctp:
nobody objects to remote backups, just peer to peer backups

dotbmp:
if Company X has some massive Node.js dependency problem that fucks all their backups on shard #113 or whatever you have your local data

bvisness:
I think peer to peer backups are overly optimistic, especially given the crap state of home internet connections in the US and the social issues of getting enough people on board

dotbmp:
I don't see how it's not a privacy problem though
prevalence of incidents is low but severity is high

bvisness:
No there are for sure privacy problems, I don’t disagree there

dotbmp:
backups are a reliability problem, not a privacy problem or a centraliztion problem.
responding to this

ctp:
if you are uploading an encrypted copy of your stuff to a server, they literally cant read it unless your key is broken

dotbmp:
good thing you have that option with the common services :stuck_out_tongue:

bvisness:
I wouldn’t want to set up this home server thing only for my bandwidth to be forever dominated by other people’s data
You need to be fairly generous to opt into a peer-to-peer scheme

ctp:
peer to peer backups would a require storage factor (tenfold, etc) as a function of the uptime and drop-in-drop-out rate of the network, and im pretty sure realistically you would require a massive factor.

dotbmp:
it's not about what's available, it's about what's convenient, and that's Google Photos and iCloud

ctp:
if remote encrypted backups were a single click in user software, the market would be a reliability-dominated competition.

dotbmp:
even a single click is too much work
people back up their photos by default
it has to be a default setting

bvisness:
My ideal would just be to have a centralized backup scheme that is all encrypted the whole way through, and with multiple providers to choose from

ctp:
^ exactly

dotbmp:
@bvisness the privacy concern means having the files encrypted before being sent

bvisness:
Yeah
I want that

Ryan:
The value of convenience to consumers is super underestimated

dotbmp:
It's everything
And it's why new software/OS paradigms that market their technical capabilities always fail
(Looking at you, Linux)
Users don't give a single fuck
You can make SuperAwesomeDevOS which is a developer's dream and it's dead in the water for the consumer market if it looks like Windows NT's inbred cousin

bvisness:
Also if stuff doesn’t run on it

dotbmp:
Or if you have to use a terminal for anything
No

bvisness:
Extremely that!!

dotbmp:
All you need is a web browser
Really
ChromeOS :smile:

Ryan:
For the average consumer this is true, though if you are targeting certain working markets you need support for existing software
e.g. no artist is going to abandon Maya just to use a slightly less awful OS

bvisness:
I think ChromeOS was a smart move

dotbmp:
Right, but if you're trying to break into the consumer market going for professionals seems like a bad starting point

Ryan:
Yeah that's a fair proposition, unless one particular professional space is easier to tackle... Like spreadsheets or financial software or whatever (no data on this, just an example)

dotbmp:
The entire content creation ecosystem is a much harder conversion than a web browser
Or more generally average consumer software
Right
I think you go for average jane consumer first, and tackle 80% solutions for low-hanging fruit

bvisness:
Unfortunately the consumer space is also where you have to compete with the big guns

dotbmp:
So is the professional space
It's all big guns

bvisness:
I guess

dotbmp:
The professional space is a lot harder to move
Lot of sunk cost
You're not gonna get the US government on HandmadeOS in the next twenty years

Ryan:
Not without force...

dotbmp:
Haha

Ryan:
#HandmadeCoup

dotbmp:
HandmadeAutocracy

bvisness:
I think the home tech space is probably the easier market at the moment

dotbmp:
I think the "home server" thing overlaps a lot with IoT which is why it's very interesting to me

Ryan:
It does seem more tractable though significant consideration should probably be afforded to various small business markets

dotbmp:
IoT is ridiculous

Ryan:
I actually have my own plan of building a home and having my own personal server, IoT style, so it's an actually smart home
And it doesn't rely on anything external

dotbmp:
Let me flick a switch on my wall which calls out to a server in California which calls back to my $200 hub which turns off my light

AsafGartner:
You'd have to wrap the home server in something else, like a game console. I don't think people would want to buy a box that doesn't seem to do anything.

dotbmp:
I already covered game console
also, it's not just a home automation server
you'd use it to power thin clients

Ryan:
Maybe media in general. Like if you have a library of movies on the server you'd be able to stream to anywhere in the house

dotbmp:
most people just need to browse the web, you could power a family of five off an underpowered server

Ryan:
Or games, Stadia-style, except locally where the speed of light is fast enough, and you can have an actual wire

dotbmp:
exactly
TVs are already just thin clients

Ryan:
Very interesting idea

dotbmp:
I mean if you're thinking in terms of grand ideas with minimal consideration for how to get there, we need digital wiring
e.g. we should have USB-C plugs in the walls instead of three-prong electric sockets
(or a better standard than USB if we're being utopian)

Ryan:
#HandmadeUtopia

dotbmp:
Indeed

Ryan:
#MarchOnIoT

dotbmp:
Right, so say you have USB-C or Thunderbolt or whatever all through your house
Just plug a display into that
Couple plugs if it needs extra bandwidth
WiFi 6 is looking pretty good for handling multiple devices

AsafGartner:
I was just going to say, you should use wireless for data.

dotbmp:
Depending on the thing
Not for displays

AsafGartner:
Have some local wireless tech in the ceiling of every room.

dotbmp:
But so, maybe you just plug in a bluetooth multiplexer by your couch

AsafGartner:
Something that doesn't go through walls that well, to avoid interference.

dotbmp:
And that's how you connect controllers
And your TV USB-Cs to the server, which has a gaming bundle

AsafGartner:
Yeah, you'd only want very few cables.

dotbmp:
The biggest practical problem obviously is acquiring the economies of scale to make this happen
And I'm sure the big boys are working on this already, but in a way that we would find repellant
(without the home server)
But the cool thing is if the home is wired for data including like, light sockets, you don't need a wifi adapter and an arm chip in every light bulb

Ryan:
You mean you don't like updating firmware on your lightbulbs?
Come on

dotbmp:
You can have colored LED smart lights that just screw in and work
Fucking
Kill me

Ryan:
:smile:
I like your thinking dotbmp, let's make it happen

dotbmp:
Yeah I'd love to be able to work on something like this

Ryan:
I meant the tech thing
Not you being killed

dotbmp:
Hahaha
HA
:rofl:
I think this would be economically viable for families aside from the upfront costs, which you have to figure, if something like this were to happen it would be a decade-long process to get a substantial segment of the market caught up
but if a Chromebook can be bought for $200 how cheap could a thin laptop be

Rory:
The problem with data over power is noise. There were Ethernet adapters which used the power main. They didn’t work well. Electrical wiring is good for delivering power not for transmission. The wiring is generally low quality.

dotbmp:
just needs a display, wifi, data/charging plug, keyboard, mouse, and a little microcontroller to power the display
@Rory this all being hypothetical it would not use the existing copper wiring
so the question would be if you were getting the stuff in the absolute most bulk manner (like giant rolls of cabling or whatever) how cheaply could you wire a house with something like USB
while being able to transmit enough power to power fridges and stuff

Rory:
The challenge is updating building codes for a higher standard 110/220v main power.
It’s possible nobody will do it until their hand is forced.

dotbmp:
hmmm
right so part of the economies of scale problem is being able to buy lobbying power :joy:
plus yeah if you're the company actually doing the rewiring you have massive legal considerations

Rory:
Yes. If you can guarantee different standards manufacturers have less risk doing what you say. They know their devices work.

dotbmp:
but I assume you'd leave that to a distributed assortment of contractors

Rory:
And the building standards we have today are just happenstance of legacy.
Going back to Edison vs Tesla.
One way is getting codes changed. Other way is make it so ubiquitous that codes follow. The second isn’t likely these days with physical things.
Like, the us still uses imperial measurements for everything. Changing power technologies is star trek level.

dotbmp:
haha well imperial -> metric would be a bigger task
anyway my take is all of this is talk, the natural starting point is a new platform (gotta make the millions somehow :^) )

Ryan:
When doing science: "The metric system is a more logically-structured and well-thought out system for defining measurements."
All other times: ":flag_us: :eagle: :red_circle: :white_circle: :blue_circle: It's about 1000 feet until the turn"

dotbmp:
which you'd distribute as a sort of shared runtime for windows/mac/linux/ios/android apps, users wouldn't know they're using this platform
then when you have a substantial dev/user base you put it on top of a microkernel
and some segment of the market will use your OS, but it will be interoperable with apps using the platform, more importantly
so there will be a network-effect-utilizing ease of transition for the most heavily invested users and devs
it leaves a bridge back to "legacy" platforms
you'd also have to pick your battles
it's very unlikely you'd get 80% of graphic design work done on your platform within 10 years
or even 20%
so you let designers use windows and macos and work on making the content creation bridge better instead
rather than trying to make a photoshop/illustrator killer
I think a new paradigm is reasonably doable if you focus on a really strong core ecosystem around what users actually care about
messaging/social media, calendars and scheduling, interoperability with other devices, media distribution and sharing

Ryan:
Hmmm that is interesting dotbmp. I see the argument but am worried about abstraction over the underlying ecosystem leaking
Like how many paradigms will de facto transfer over, even if we don't want them to?

dotbmp:
yeah it's a big concern
and if you go for a no-compromises position, you're sacrificing a lot of efficiency

Ryan:
It would maybe be important to understand the vision for the custom hardware (and everything on it that would come later, like an OS) very early so that abstraction over existing infrastructure is done more responsibly

dotbmp:
yeah, and striking a balance between idealism and practicality
like, casey's standardized ring buffer I/O idea is great imo
but realistically you have to support USB
even if you had the ability to push a new, simplified standard, you need legacy support
I think it's probably reasonable to use RISC-V as a target architecture
it's not perfect but it's very simple compared to the alternatives, open, about 1/10th the IP cost of ARM, and obviously a big up and comer gaining a lot of investment right now
and as long as you have the CPU architecture in mind, that's pretty much most of the hardware consideration in terms of system design
is miotatsu here haha cause I would hate to reiterate this stuff on the forums :rofl:

bvisness:
probably not
better start copy-pasting :)

dotbmp:
:rip: