Slashdot Mirror
Free Software Foundation Shakes Up Its List of Priority Projects (networkworld.com)
alphadogg quotes Network World:
The Free Software Foundation Tuesday announced a major rethinking of the software projects that it supports, putting top priority on a free mobile operating system, accessibility, and driver development, among other areas. The foundation has maintained the High Priority Projects list since 2005, when it contained just four free software projects. [That rose to 12 projects by 2008, though the changelog shows at least seven projects have since been removed.] Today's version mostly identifies priority areas, along with a few specific projects in key areas.
The new list shows the FSF will continue financially supporting Replicant, their free version of Android, and they're also still supporting projects to create a free software replacement for Skype with real-time voice and video capabilities. But they're now also prioritizing various projects to replace Siri, Google Now, Alexa, and Cortana with a free-software personal assistant, which they view as "crucial to preserving users' control over their technology and data while still giving them the benefits such software has for many."
And other priorities now include internationalization, accessibility, decentralization and self-hosting, and encouraging governments to adopt free software.
The new list shows the FSF will continue financially supporting Replicant, their free version of Android, and they're also still supporting projects to create a free software replacement for Skype with real-time voice and video capabilities. But they're now also prioritizing various projects to replace Siri, Google Now, Alexa, and Cortana with a free-software personal assistant, which they view as "crucial to preserving users' control over their technology and data while still giving them the benefits such software has for many."
And other priorities now include internationalization, accessibility, decentralization and self-hosting, and encouraging governments to adopt free software.
"We can't build a kernel in a reasonable amount of time, so instead we'll take on projects that meld AI, natural language and voice recognition!!"
"I don't know, therefore Aliens" Wafflebox1
I looked at the projects from 2008, and AFAIK none of them have really made much progress or are already dead. Gnash, really? Luckily for everyone (except Adobe) Flash is in it's death knell in 2017. Coreboot is a great concept (I have used LinuxBIOS in a couple of projects) but ultimately doomed because firmware/BIOS is intimately tied to hardware - it will be great for hobby projects but by definition never be useful for mainstream PCs. And so it goes down the list...
And the 2017 list... Free smartphone OS basically seems to be "free Android" - I'm sure it will be about as successful at the 2008 goal with "gNewSense". FSF personal assistant? Could it be possible they don't understand how these work? It's trivial client software with billions of dollars in server hardware behind it. And seriously, "projects that replace Google, Facebook Apple, and so on"? Again, you don't replace those unless you have billions in backend investment and billions of users.
I commend them on finally trying to address the totally dysfunctional open source community in terms of female and minority inclusion, but they still need to prioritize actual useful *projects*, not just processes...
I read TFA because I was curious to see what the FSF felt they'd accomplished in the last twelve years; but, alas, there's no mention of the list's former contents.
I have trouble believing a group which has been unable to get a working Hurd released sometime during the past three decades is capable of accomplishing any of their stated "high priority" goals.
I don't mean to discount their philosophical importance; but really I think that's pretty much the sum total of their impact.
#DeleteChrome
Linux is not a GNU project. That's why not.
Free software assistant... already exists
http://mycroft.ai
They've got an RPi image you can download, slap on a card, and be up and running with a USB mic and something to handle the audio out.
Seems to me like the FSF should pay more attention to what is already going on.
I've fallen off your lawn, and I can't get up.
And the 2017 list... Free smartphone OS basically seems to be "free Android" - I'm sure it will be about as successful at the 2008 goal with "gNewSense". FSF personal assistant? Could it be possible they don't understand how these work? It's trivial client software with billions of dollars in server hardware behind it. And seriously, "projects that replace Google, Facebook Apple, and so on"? Again, you don't replace those unless you have billions in backend investment and billions of users.
Originally "intelligent agents" were supposed to be software running on the users behalf which strived to understand contextually what the users needs were and act to support the user. What actually happened big data cyber stalking firms have entirely corrupted this vision.
It isn't that you start over and run parallel infrastructure. It is more about designing local agents able to effectively leverage the network as it is to fulfill needs of the user. There is no reason an IA can't run a google, wolfram alpha, bing search, check specialized databases of interest or rummage through your email or local files on your behalf. The only difference is the IA is acting in YOUR best interests NOT a third parties and it isn't sending all of your local personal shit to god knows who for god knows why.
Current systems are more than capable of doing NLP and voice recognition locally. Even if you go with generic ANN approach for recognition you don't need exotic hardware to use a trained network. Granted all of this requires specialized skills but far from unreasonable.
If done properly you can provide value with IA's the likes of siri, cortana, alexa...etc can't because it's not in their business model. If you got the basic interfaces, perhaps some specialized DSLs and focus on making it easy for people to build their own agent logic and share it with others.. there is a chance... perhaps a small one of creating something that snowballs where the value and the capabilities of the IA grows organically as more people contribute or improve upon logic that scratches their itches.
Do you just not know anything about this very publicly discussed topic, or are you one of the douchebags who actively discourages accepting commits from women or others outside of your social circle?
You're still going to need a MITM unless everyone figures out how to do proper port forwarding or exposing the port from behind the ISP's modem's router/firewall.
Apparently, Linux people still fail at understanding some of these basic networking concepts. Not surprised.
All you need is something that looks like a naming/directory/oldschool ILS type server to coordinate things.
With IPv6 implementations ports are mapped 1:1 because there is no packet mangling. All you need to do is prime the SPI with a few UDP packets in either direction and your good to go.
With IPv4 NATs ... especially the CGN variety without a 1:1 map required much more creative approaches such as birthday paradox spam across the port space which is slow and unreliable and may not work at all due to lack of common overlap in mapped space.
I use owncloud and I really love it. It's been awesome to be able to access my files from anywhere while still maintaining control of them down to the metal. I sync my calendar and contacts with my phone through caldav and carddav. I am very close to running cyanogenmod/lineages without google play services. It certainly isn't as convenient as Google, but freedom isn't free. I am glad that I can get a little bit of control back.
Current systems are more than capable of doing NLP and voice recognition locally. Even if you go with generic ANN approach for recognition you don't need exotic hardware to use a trained network. Granted all of this requires specialized skills but far from unreasonable.
Since you seem to know the basics of ANN-based AI but not the details, check this article out to get into the current decade, it's a good overview of how much resources it really takes to do ANN right: https://www.nytimes.com/2016/1...
Summary is: sure you can do really crappy NLP locally, but what Google has started doing is at another level entirely. And that's not even at the level that will be required to really get "intelligent agents" to be truly useful. The limiting factor to ANN right now is computing power (and/or much more special purpose hardware). That's the reason it was nearly abandoned 20 years ago and revived recently - it was impossible with the previous concepts of "supercomputers", etc and now only with massively distributed computing has it been possible to actually simulate the multilevel networks required.
Chromebooks use Coreboot, and they regularly top the lists of most popular laptops on Amazon. The firmware/binary blog thing isn't as much of an issue as you might think, since the basic idea of Coreboot is to do the minimum possible to boot the OS rather than replace all the random BIOS functions and crud built up over the years.
Replicant is likely a response to Cyanogen giving up, and an attempt to find some way around the binary blob hell that is smartphone chipsets. Well, these days Android runs on a lot more than just phones, and things like tablets tend to have more transparent hardware for their radios so it's far from an impossible goal.
Personal assistants could easily use your own personal server. The speech recognition might not be quite as good, but of course you can just type stuff. In any case, being able to look up results on your choice of search engine or Wikipedia, and being able to interpret simple commands like "set a reminder for next Tuesday at 7 PM" hardly requires billions of dollars of hardware. There are some useful Google Now features I don't use for privacy reasons, like traffic info cards, which could easily be replaced by free software even if I have to explicitly tell it my route home from work rather than it using machine learning to figure it out for me.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
I welcome your instructions on how to ring people's telephones in other countries while I'm myself travelling internationally without using either a telephone or Skype to do this, at a cost comparable to what I'd spend using SkypeOut. Preferably legally. Thanks.
Il n'y a pas de Planet B.
Again, you don't replace those unless you have billions in backend investment and billions of users.
There are some really interest projects going on that allow for decentralization of applications (eg ipfs, blockchain technology). With the continued increase in broadband speeds and distributed technologies it's not as far fetched as it once was.
given the opportunities for snooping, they level of insecurity that is here and only going to get worse, a free mobile OS would be a great thing.
google's efforts at completely eliminating your privacy (i HAVE to sync my calendar to the google, WTF?!) is evil.
couple of problems:
how does this work with the carriers. are they free to keep your phone off the air once they start doing OS checks ?
pushing an OS onto a variety of hardware, just as on PCs is definitely painful, and I think it's going to be much worse for mobile phones. Lots of parts are proprietary and require NDAs to have access to datasheets. Not sure how you get around this problem.
Absolute statements are never true