|
|
07-08-2019, 10:36 PM
|
|
mesospheric bore
|
|
Join Date: Jul 2005
Location: New Zealand
Gender: Male
|
|
Re: Privacy, Anonymity, and Compartmentalization
|
07-09-2019, 01:32 AM
|
|
Shitpost Sommelier
|
|
|
|
Re: Privacy, Anonymity, and Compartmentalization
Welp. Time to get me some Faygo.
__________________
Peering from the top of Mount Stupid
|
07-09-2019, 03:07 AM
|
|
I read some of your foolish scree, then just skimmed the rest.
|
|
Join Date: Jan 2005
Location: Bay Area
Gender: Male
|
|
Re: Privacy, Anonymity, and Compartmentalization
The facial detection software used for years in things like phones and cameras is a pretty simple algorithm that hunts the image specifically for the light and dark contrast created by the T of your eyes and nose and then expands out from there to refine the search making the assumption that once it’s found the T, it should only be so far before it finds the A, er I mean the chin, or similar features. If it finds that contrast it assumes face and puts a box around it. It’s a simple algorithm and it works way better than anyone expected for how dumb it is.
The article mentions something like face ID, but that isn’t what this makeup would be for. Instead if those building facial recognition software take any shortcuts, like say decides to use this algorithm to toss out bits of the image or video it doesn’t think is a face so the big processor intensive facial recognition guns can only focus on relevant data then a bit of makeup will go a long way to keeping the machine from even assuming you have a face to begin with.
This all does make me wonder what exactly face recognition engineers are doing with people who wear photo realistic t-shirts of other people.
|
07-09-2019, 04:04 AM
|
|
Shitpost Sommelier
|
|
|
|
Re: Privacy, Anonymity, and Compartmentalization
Quote:
Originally Posted by Ari
This all does make me wonder what exactly face recognition engineers are doing with people who wear photo realistic t-shirts of other people.
|
While I get that it could be a problem that lets criminals into houses from Facebook photos:
Quote:
Originally Posted by Kamilah Hauptmann
|
__________________
Peering from the top of Mount Stupid
|
07-09-2019, 06:24 AM
|
|
mesospheric bore
|
|
Join Date: Jul 2005
Location: New Zealand
Gender: Male
|
|
Re: Privacy, Anonymity, and Compartmentalization
Quote:
Originally Posted by Ari
The facial detection software used for years in things like phones and cameras is a pretty simple algorithm that hunts the image specifically for the light and dark contrast created by the T of your eyes and nose and then expands out from there to refine the search making the assumption that once it’s found the T, it should only be so far before it finds the A, er I mean the chin, or similar features. If it finds that contrast it assumes face and puts a box around it. It’s a simple algorithm and it works way better than anyone expected for how dumb it is.
The article mentions something like face ID, but that isn’t what this makeup would be for. Instead if those building facial recognition software take any shortcuts, like say decides to use this algorithm to toss out bits of the image or video it doesn’t think is a face so the big processor intensive facial recognition guns can only focus on relevant data then a bit of makeup will go a long way to keeping the machine from even assuming you have a face to begin with.
This all does make me wonder what exactly face recognition engineers are doing with people who wear photo realistic t-shirts of other people.
|
I get the impression that various image recognition techniques are quite easy to fool. iNaturalist can be surprisingly good at getting the right genus of a photo - if there's a living thing in the pic. But last time I checked it was pretty determined to match something and didn't have any "not a living thing" results.
Also, see this: Deep neural networks are easily fooled: High confidence predictions for unrecognizable images | Evolving AI Lab
|
04-10-2020, 10:49 PM
|
|
Solipsist
|
|
Join Date: Jul 2004
Location: Kolmannessa kerroksessa
Gender: Male
|
|
Re: Privacy, Anonymity, and Compartmentalization
Relevant here too.
Quote:
Originally Posted by JoeP
Coronavirus: Apple and Google team up to contact trace Covid-19 - BBC News
Quote:
But ultimately, they aim to do away with the need to download dedicated apps, to encourage the practice.
The two companies believe their approach - designed to keep users, whose participation would be voluntary, anonymous - addresses privacy concerns.
Their contact-tracing method would work by using a smartphone's Bluetooth signals to determine to whom the owner had recently been in proximity for long enough to have established contagion a risk.
If one of those people later tested positive for the Covid-19 virus, a warning would be sent to the original handset owner.
No GPS location data or personal information would be recorded.
|
No need to download dedicated apps = no user control, right?
No personal information would be recorded, yet the point is to contact the original handset owner: how can they do that without personal information?
Quote:
"Privacy, transparency and consent are of utmost importance in this effort and we look forward to building this functionality in consultation with interested stakeholders," Apple and Google said in a joint statement.
|
Someone is lying here.
|
|
04-10-2020, 10:53 PM
|
|
California Sober
|
|
Join Date: Jul 2004
Location: Silicon Valley
Gender: Bender
|
|
Re: Privacy, Anonymity, and Compartmentalization
How many fuckin do you need?!
|
04-10-2020, 11:31 PM
|
|
Solipsist
|
|
Join Date: Jul 2004
Location: Kolmannessa kerroksessa
Gender: Male
|
|
Re: Privacy, Anonymity, and Compartmentalization
ALL THE THANKS.
|
04-10-2020, 11:32 PM
|
|
Solipsist
|
|
Join Date: Jul 2004
Location: Kolmannessa kerroksessa
Gender: Male
|
|
Re: Privacy, Anonymity, and Compartmentalization
Actually thanksgrabbing was not my motivation for crossposting - merely an added bonus.
|
02-21-2021, 09:18 PM
|
|
Solitary, poor, nasty, brutish, and short
|
|
|
|
Re: Privacy, Anonymity, and Compartmentalization
This article does a much better job of explaining this than I can, but it is really really important for people to understand. So you should read it.
A Case Against the Peeping Tom Theory of Privacy | WIRED
|
02-21-2021, 11:00 PM
|
|
Solipsist
|
|
Join Date: Jul 2004
Location: Kolmannessa kerroksessa
Gender: Male
|
|
Re: Privacy, Anonymity, and Compartmentalization
Um, so, I read it, and just as I thought they were about to get into the actual meat of what these non-peeping-Tom non-focus-on-the-individual privacy threats actually were, I got to ... the end.
I need you to do a better job of explaining it that that article did, even though you claim you can't.
|
02-22-2021, 12:30 AM
|
|
Solitary, poor, nasty, brutish, and short
|
|
|
|
Re: Privacy, Anonymity, and Compartmentalization
In terms of microtargeting, companies are using algorithms trained on other people's interactions with media to guide them toward increasingly radicalized content, because that is what keeps people engaged.
So that's bad in and of itself. Outrage increases user engagement, so any algorithm meant to keep people online more will guide people toward increasingly outrageous content.
But because it's only being shown to receptive audiences, other people literally don't even know this stuff is out there in order to challenge or rebut it. This happens a lot with straight up political ads, but also with other more nebulous ideas like anti-vax and Qanon. Think about how these things seem to pop out seemingly out of nowhere with throngs of fully radicalized true believers. That's because those groups were chugging along indoctrinating people sometimes for years. But people who were not inclined toward that type of thing never even saw it until it reached the point that it got loud and publicly violent.
On top of that, a lot of it is used for redlining. If you just set a computer out to learn about real people in the real world with all its systemic injustices, it will codify those injustices. I used to work with some narrow AIs designed to predict infrastructure demands. Fortunately, the companies I did it with were heavily regulated and being watched pretty closely, so we had to have a lot of human intervention with these learning systems because they would always predict that poor and minority areas didn't require as advanced an infrastructure as wealthy white areas because that's the way it'd always been. But most industries and companies aren't regulated at all, and their targeting methods aren't done out in the open. But they are happening, and people don't know how their internet looks different from anyone else's. And because these choices are made by these impenetrable algorithms, you can't really prove that's what's happening, even when it clearly is.
|
02-22-2021, 01:46 AM
|
|
I read some of your foolish scree, then just skimmed the rest.
|
|
Join Date: Jan 2005
Location: Bay Area
Gender: Male
|
|
Re: Privacy, Anonymity, and Compartmentalization
Quote:
Originally Posted by lisarea
But because it's only being shown to receptive audiences, other people literally don't even know this stuff is out there in order to challenge or rebut it.
|
This!
Reading through it I couldn’t help but think of my youtube ads. I’ve been getting down right scam ads on youtube videos, including, I kidd you not, Ball Enlargement supplements. After looking at my ad info to see wtf, it turns out I’m not tracked, and thus am worth less to google (and no, I don’t think I’m completely untracked). Youtube gives me mostly bargain basement filler ads, and I’m sure those who are tracked better don’t realize just how low youtube will accept. There’s easily a possibility for class disparity where if your likes and dislikes are considered worth less monetarily than others, you get fed more scams and junk.
|
02-22-2021, 04:18 AM
|
|
Shitpost Sommelier
|
|
|
|
Re: Privacy, Anonymity, and Compartmentalization
Bing, which I don't use very often, for just about any video search handed me from a blank slate a pile of far right news site videos.
__________________
Peering from the top of Mount Stupid
|
02-22-2021, 06:03 PM
|
|
Adequately Crumbulent
|
|
Join Date: Jan 2005
Location: Cascadia
Gender: Male
|
|
Re: Privacy, Anonymity, and Compartmentalization
Hmm, so about those ball enlargement supplements...
|
04-14-2021, 10:00 PM
|
|
Solitary, poor, nasty, brutish, and short
|
|
|
|
Re: Privacy, Anonymity, and Compartmentalization
Data Brokers Are a Threat to Democracy | WIRED
I love it when I see someone bring this up in any remotely mainstream media. It's one of those things where I'm sitting here waiting for other people to notice it. I talk about it, as you all are painfully aware, but pretty much all I can do is contact my representatives and complain on the internet. And I mostly get Cassandraed about it when I do, so fat lot of good that does.
So please listen to that guy named Justin Sherman. He is an important person who writes for Wired.
|
04-14-2021, 11:52 PM
|
|
Solipsist
|
|
Join Date: Jul 2004
Location: Kolmannessa kerroksessa
Gender: Male
|
|
Re: Privacy, Anonymity, and Compartmentalization
Interesting title choice. He mentions some threat to national security and the possibility of foreign powers influencing elections.
But the threat to civil rights is much more significant and direct, and he deals with it first and has plenty of examples of it. Why not lead with that in the heading? Is civil rights a dirty (commie, libruhl) expression in the US?
|
04-14-2021, 11:56 PM
|
|
Adequately Crumbulent
|
|
Join Date: Jan 2005
Location: Cascadia
Gender: Male
|
|
Re: Privacy, Anonymity, and Compartmentalization
The author doesn't usually pick the title. I'm guessing WIRED thought "threat to democracy" would get more clicks.
|
04-15-2021, 12:23 AM
|
|
Solitary, poor, nasty, brutish, and short
|
|
|
|
Re: Privacy, Anonymity, and Compartmentalization
That's what I figure. And threats to civil liberties could reasonably be framed as threats to democracy, too, so it's not a terrible umbrella term for all that business.
|
04-15-2021, 10:27 AM
|
|
Solipsist
|
|
Join Date: Jul 2004
Location: Kolmannessa kerroksessa
Gender: Male
|
|
Re: Privacy, Anonymity, and Compartmentalization
Oh for sure. I wasn't knocking the author. Just ... increasing my post count. For democracy!
|
04-15-2021, 06:03 PM
|
|
A Very Gentle Bort
|
|
Join Date: Jan 2005
Location: Bortlandia
Gender: Male
|
|
Re: Privacy, Anonymity, and Compartmentalization
__________________
\V/_ I COVLD TEACh YOV BVT I MVST LEVY A FEE
|
05-05-2021, 12:07 AM
|
|
Solitary, poor, nasty, brutish, and short
|
|
|
|
Re: Privacy, Anonymity, and Compartmentalization
This thread, I guess?
Colorado bill prohibits insurers from using “discriminatory” data, like social media and credit scores, to set rates
I would much prefer insurers were prohibited from using that sort of information from data brokers at all. In fact, I'd rather they took a prescriptive approach, and limited them to using a short list of specific, verified information that they have explicit reasons for including. Nothing else.
But I would LOVE to see what they're using, and how the regulators would go about analyzing it. It's usually a pretty opaque process without a lot of human intervention in the initial stages, so it would probably take a bit of detective work to tangle out actionable discrimination.
|
05-31-2021, 05:16 PM
|
|
Solitary, poor, nasty, brutish, and short
|
|
|
|
Re: Privacy, Anonymity, and Compartmentalization
People using Luddite to mean technophobe is a longstanding peeve of mine, so this is of interest to me.
It's really long, though. Do you have maybe time markers for especially good parts?
|
06-01-2021, 05:08 PM
|
|
Admin
|
|
Join Date: Apr 2004
Location: Ypsilanti, Mi
Gender: Male
|
|
Re: Privacy, Anonymity, and Compartmentalization
Oh sorry! The actual interview is only from about 24:00 to 1:05:00.
|
Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
|
|
Thread Tools |
|
Display Modes |
Linear Mode
|
Posting Rules
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off
|
|
|
All times are GMT +1. The time now is 08:17 AM.
|
|
|
|