[00:00.000 --> 00:00.400]  Hi, Devon.
[00:01.720 --> 00:12.300]  So, as it says, I'm Alex Matthews, National Chair of Restore the Fourth, which is a troublemaking civil liberties organisation that opposes mass government surveillance.
[00:13.080 --> 00:16.840]  And we have some of our Restore the Fourth folks here in the audience.
[00:17.020 --> 00:17.840]  Shout out to you guys.
[00:18.480 --> 00:23.080]  So, this talk is on surveillance give me chills.
[00:24.480 --> 00:35.840]  It relates to some research that I and some other researchers have been doing to explore and to quantify the chilling effect that arises from surveillance.
[00:36.800 --> 00:48.500]  One of the biggest questions in the surveillance world is the fundamental question that we get asked all the time, which is, do people actually care?
[00:48.760 --> 00:53.720]  And the research that I'll be talking about today is an attempt to get at a part of that question.
[00:53.980 --> 00:58.120]  So, part of the preconception is that maybe people don't care.
[00:58.120 --> 01:00.200]  Maybe people are ignorant.
[01:00.400 --> 01:02.460]  Maybe people are apathetic.
[01:02.660 --> 01:05.380]  Maybe people have other shit to deal with, frankly.
[01:07.420 --> 01:32.340]  And particularly, there is a stereotype about young people like Becky with the good hair over here that maybe they just lay it all out online and maybe we can interpret that as being that they are happy to have law enforcement have access to the data that they put online and use it against them.
[01:32.340 --> 01:37.540]  So, I am more than a little bit sceptical of this stereotype.
[01:38.600 --> 01:57.240]  But when I started off this research project, which was immediately after the Snowden revelations in 2013, I actually thought it was very possible that what we would come up with was that there would not be a measurable chilling effect from surveillance.
[01:57.240 --> 02:02.960]  And it was something that researchers had had a lot of problems in identifying and quantifying before.
[02:03.380 --> 02:14.580]  The advantage of the Snowden revelations is that they provided what social science researchers call an exogenous shock that enables you to measure more precisely what actually happened.
[02:14.580 --> 02:34.720]  The first element of this that I am going to talk about is the DHS list, which is a list of terms from the Department of Homeland Security Social Media Monitoring Unit, which is a thing, as to what they are interested in people posting about online.
[02:35.240 --> 02:45.940]  Then we will go into a look at the recent research and then into the cases where social media postings have triggered actual law enforcement involvement.
[02:46.440 --> 03:00.960]  We will be talking about some of the ways in which governments define and redefine or choose not to define the term violent extremist and how they go about suppressing what they define as violent extremist content online.
[03:01.640 --> 03:08.460]  And then I will be offering some very limited and provisional non-state-based solutions to the problem.
[03:10.600 --> 03:22.180]  So one of the most interesting questions that is a part of this is how do we know what the government is looking for when it comes to social media, when it comes to Google searches?
[03:23.500 --> 03:37.700]  And the DHS list, which dates from 2011 and was definitely still in use still in mid-2013, gives a limited take from one agency on what they are actually looking for.
[03:37.700 --> 03:40.540]  So we are going to play a game.
[03:40.980 --> 03:45.180]  It is a very, very, very simple game compared to most of the games here at the conference.
[03:45.560 --> 03:50.560]  What I am going to do is I am going to present you with six pairs of terms.
[03:51.100 --> 03:59.640]  One of them is on the DHS list and therefore is something that the government is looking for in terms of postings online and one of them is not.
[04:00.140 --> 04:07.460]  Keep track of what you think the right answer is and then I am going to be asking at the end of the six terms who got them all right.
[04:09.480 --> 04:11.420]  So let's start off...
[04:13.320 --> 04:15.340]  Man, I hope it doesn't fade into the next screen.
[04:17.160 --> 04:18.520]  Here we have two terms.
[04:20.640 --> 04:21.880]  Agriculture and weed.
[04:23.060 --> 04:26.360]  One of these is on the DHS list and one of them is not.
[04:26.880 --> 04:30.760]  Take a moment to think about which it is and then I am going to show you the answer.
[04:30.760 --> 04:32.420]  Three, two, one.
[04:35.060 --> 04:36.620]  The answer is agriculture.
[04:37.980 --> 04:47.280]  We think that perhaps this is because the DHS is concerned, among other things, with eco-terrorism and with disruptions to the food supply.
[04:47.660 --> 04:54.040]  This does not mean, of course, that the federal government as a whole is unconcerned about weed because evidently it is not.
[04:54.180 --> 05:02.420]  It just means that in the context of the DHS social media monitoring unit, they choose not to use that term to search for suspicious activity.
[05:03.140 --> 05:04.140]  So, next one.
[05:09.270 --> 05:14.320]  More conventional stuff that gets talked about in terms of what the surveillance state may be looking for online.
[05:14.560 --> 05:16.680]  And one of these terms is on there, one of them is not.
[05:16.940 --> 05:17.960]  Take a moment to think about it.
[05:18.270 --> 05:19.960]  Three, two, one.
[05:22.470 --> 05:26.650]  And what they are looking for, this perhaps is the least surprising of the six, is jihad.
[05:26.650 --> 05:44.220]  But it's interesting in the context of the shootings in San Bernardino, where the government actually came under fire for not having been interested in or having flagged sufficiently postings of the San Bernardino shooters on the topic of jihad.
[05:44.520 --> 05:49.060]  And the interesting thing is here that jihad as a search term is actually on a government list.
[05:49.060 --> 06:06.420]  It's just that presumably the volume of postings on the topic of jihad, the vast majority of which will be innocent, is so large that in practice it was not possible to drill down and identify these things ahead of time, which is what we run into as a problem a lot in mass surveillance systems.
[06:06.420 --> 06:08.500]  So, next one.
[06:11.490 --> 06:15.050]  White power and illegal immigrants.
[06:16.490 --> 06:18.130]  Take a moment to think about it.
[06:19.050 --> 06:20.710]  Three, two, one.
[06:23.460 --> 06:26.260]  And it is illegal immigrants that is on the list.
[06:26.420 --> 06:27.560]  And white power is not.
[06:27.900 --> 06:37.820]  We think that this is because the DHS has something of a focus on cartels smuggling people across the border and wanting to disrupt them.
[06:38.240 --> 06:50.320]  But this points up an interesting aspect of the list that the terms included on it are terms from the analyst's perspective of topics that they might be interested in.
[06:50.480 --> 06:55.800]  And then there isn't always what looks like careful thought going into them as to what people might actually post about.
[06:56.040 --> 07:04.680]  Because if you're a coyote trying to smuggle people across the border, are you really going to post the term illegal immigrants on Facebook or Twitter?
[07:04.680 --> 07:06.560]  Well, probably not.
[07:08.320 --> 07:09.040]  Next one.
[07:12.940 --> 07:14.140]  Hacktivist and anonymous.
[07:14.380 --> 07:16.600]  And, disappointingly, one of them is on the list.
[07:17.180 --> 07:20.480]  So, three, two, one.
[07:20.660 --> 07:21.400]  Which of them is it?
[07:22.500 --> 07:24.100]  Then the answer is hacktivist.
[07:24.400 --> 07:30.160]  But, fortunately, absolutely nobody in this room would ever post on social media on the topic of hacktivism.
[07:30.280 --> 07:30.920]  So, we're all okay.
[07:31.100 --> 07:31.480]  Am I right?
[07:32.520 --> 07:33.220]  Thank you.
[07:34.160 --> 07:36.240]  I'm glad to be with such non-suspicious people.
[07:36.420 --> 07:37.080]  It's so comforting.
[07:40.590 --> 07:41.110]  Next one.
[07:42.330 --> 07:45.150]  Another couple of topics that may be of interest to some people in the room.
[07:45.250 --> 07:46.150]  One of these is on the list.
[07:46.250 --> 07:46.890]  One of them is not.
[07:49.150 --> 07:52.910]  BitTorrent and freaking spelt with a PH.
[07:56.770 --> 08:02.350]  And, yes, it looks like the Department of Homeland Security is very interested in whether you get your free card.
[08:06.150 --> 08:06.870]  Last one.
[08:10.420 --> 08:12.740]  Zero-day and revolution.
[08:13.620 --> 08:14.440]  One of them on the list.
[08:14.500 --> 08:15.100]  One of them is not.
[08:16.560 --> 08:18.480]  Three, two, one.
[08:20.940 --> 08:22.620]  And they're interested in zero day.
[08:23.580 --> 08:29.460]  This is interesting principally because it's not as if violent revolution might not be a concern of the governments.
[08:31.640 --> 08:36.140]  But maybe they felt that revolution would be too ambiguous a term to search on.
[08:36.300 --> 08:40.740]  After all, as we all know now, Bernie Sanders preaches a revolution of a kind.
[08:41.460 --> 08:43.920]  And maybe they wanted to exclude that sort of thing.
[08:44.180 --> 08:47.260]  In any case, zero day vulnerabilities seem to be a concern.
[08:48.420 --> 08:51.380]  So, who in the audience got four of the six right?
[08:52.600 --> 08:53.120]  Ooh!
[08:54.120 --> 08:55.800]  Who got five of the six right?
[08:56.960 --> 08:59.200]  Okay, and who got six of the six right?
[09:01.360 --> 09:02.680]  Absolutely goddamn nobody!
[09:05.220 --> 09:20.140]  What this illustrates is that even for an expert audience, like you guys are, who are knowledgeable in this, and who are interested enough in the topic of social media surveillance, to come and listen even to me about social media surveillance.
[09:21.140 --> 09:23.180]  Nobody got six out of six right.
[09:23.960 --> 09:30.680]  And this poses interesting questions as to the legibility of these systems to the general public.
[09:30.860 --> 09:45.880]  If we don't know, if we cannot tell what the government is really interested in, then there's going to be a mismatch and an inability for us to tailor our behavior or tailor our searches to deal with the phenomenon of mass government surveillance.
[09:46.500 --> 10:00.680]  And that, to an extent, is going to be intentional, because government agencies frequently express that they do not want people to be trying to evade mass surveillance systems by tailoring what they do and do express online.
[10:00.680 --> 10:15.060]  So, with that in mind, that there is that mismatch there, we're going to go into three research papers that are recent papers that are worth taking a look at in this context.
[10:17.220 --> 10:24.720]  So, for our paper, we were asking a more detailed version of Do People Care?
[10:24.720 --> 10:28.960]  which is Does Knowledge of Mass Surveillance Lead to Self-Censorship Online?
[10:29.360 --> 10:37.040]  The title of the paper that I co-authored with Professor Catherine Tucker at MIT is Government Surveillance and Internet Search Behavior.
[10:38.160 --> 10:47.020]  And we looked at how search volume on Google Trends changed after versus before the PRISM revelations.
[10:47.020 --> 11:12.540]  And for people who don't know the term PRISM, that is the specific revelation on June 6th of 2013, that where the NSA was depicting that they had partnerships with major tech companies, Google, Yahoo, Microsoft, Skype, Dropbox, a few others, to have access to their data in bulk.
[11:12.540 --> 11:25.620]  There were arguments and confusion over what this actually entailed, but there have been reports that, say, 93% of U.S. government surveillance collection is classed under the PRISM program, so it's a major thing.
[11:26.100 --> 11:35.140]  We looked at Google Trends data for 100,000-plus observations of specific search engine search terms.
[11:36.400 --> 11:46.220]  We developed a list of 282 search terms, and a lot of this analysis is going to be contingent on whether these search terms can be considered to be meaningful for the analysis.
[11:47.440 --> 11:50.800]  So, I'm going to talk a little bit about how we came up with those lists.
[11:51.000 --> 11:54.640]  So, we started off with this DHS list of sensitive search terms.
[11:55.140 --> 11:58.180]  I have the file of the search terms on my computer.
[11:58.400 --> 11:59.840]  People can take a look at that.
[11:59.960 --> 12:00.860]  I can give you a link.
[12:01.740 --> 12:03.040]  But we also needed more.
[12:03.280 --> 12:11.620]  We needed a control list of search terms that were unambiguously innocent, that would not get you in trouble with government, that would not be personally embarrassing either.
[12:11.620 --> 12:26.700]  And for that, we used the Google Zeitgeist list that is published every year, where Google lists 100 search terms that people are using a lot online as a way of analyzing shifts in culture.
[12:26.940 --> 12:34.840]  So, this list contains names of celebrities, names of people who die, major sporting events, new launches of tech products.
[12:34.840 --> 12:38.400]  It contains absolutely no references to porn whatsoever.
[12:39.280 --> 12:41.840]  So, there's certainly some curation going on.
[12:41.980 --> 12:49.140]  But it is some reflection of what is happening in a given year on the clean Internet.
[12:51.140 --> 12:53.720]  So, it serves as a good control for our purposes.
[12:54.360 --> 12:59.160]  Except for term number 93 on the list, which was Edward Snowden, so we only used the top 50.
[13:00.940 --> 13:18.360]  So, then we also needed, because we wanted to tease out the differences between stuff that people thought would get them in trouble with the government and stuff that they would think was personally embarrassing, we crowdsourced at a tech incubator a list of 100 personally embarrassing terms.
[13:18.380 --> 13:21.160]  And I'll talk a little bit later about what those may have included.
[13:22.800 --> 13:27.060]  But, even so, think back to what we were doing with that exercise just now.
[13:27.320 --> 13:37.020]  There is a mismatch between what the government thinks you will post about that will get you in trouble and what the general public thinks will get them in trouble for posting about.
[13:37.300 --> 13:49.600]  So, we still needed to independently rate the terms on these lists using Amazon Mechanical Turk workers for how likely this term is to get me in trouble with the U.S. government and how likely it is to get me in trouble with my family.
[13:50.200 --> 13:53.380]  So, these are the findings that we came up with.
[13:54.580 --> 14:02.780]  A small but precisely measured chilling effect in the U.S. of about 5% on search volume for government sensitive terms.
[14:03.000 --> 14:06.840]  This is like anthrax and pipe bomb and a few other things.
[14:07.560 --> 14:15.620]  And note, within this study, we are not making a judgement as to whether suppressing that search volume is a good or a bad thing.
[14:15.740 --> 14:18.060]  We are just reflecting that it is happening.
[14:19.040 --> 14:28.600]  In non-U.S. countries among the top 10 U.S. trading partners, there is also a chilling effect on personally sensitive terms.
[14:28.760 --> 14:34.320]  And there, we are not only talking about the medical things listed there like eating disorder, depression, erectile dysfunction.
[14:34.600 --> 14:39.760]  We are also talking about other things that may be embarrassing like My Little Pony, Nickelback and LARPing.
[14:43.450 --> 14:53.990]  And, to be honest, when we went out and validated this and tried to find what people thought would be embarrassing terms that wouldn't get them in trouble with the government, so many people mentioned My Little Pony.
[14:53.990 --> 14:54.670]  It was freaky.
[14:55.530 --> 14:58.010]  And also foot fungus of various kinds.
[14:58.170 --> 14:58.550]  It was gross.
[15:01.050 --> 15:06.690]  So, this was the first study to actually put a number on what was happening after the Snowden revelations.
[15:06.830 --> 15:16.890]  Because there have been surveys where entities like Pew Research and Penn, which is a writer's organisation, had gone out and said, we are calling you about the Snowden revelations.
[15:16.890 --> 15:18.610]  Do you think these are very serious?
[15:18.710 --> 15:19.990]  Yes, I think they're very serious.
[15:20.430 --> 15:22.530]  Have you taken any actions in response to them?
[15:22.610 --> 15:23.450]  Has it changed your behaviour?
[15:23.590 --> 15:24.490]  Yes, yes, of course.
[15:25.570 --> 15:28.790]  We were sceptical of this because of social desirability bias.
[15:29.210 --> 15:38.070]  If you've said that the Snowden revelations are a serious thing, then you might say that you are changing your behaviour when you haven't actually done squat, not even changing a password.
[15:39.050 --> 15:43.270]  So, we wanted to get at actual user behaviour and this was the first study that did.
[15:43.270 --> 15:45.270]  And as a result, made something of a splash.
[15:45.730 --> 15:49.510]  Most downloaded paper in July 2014 on SSRN.
[15:50.050 --> 15:54.310]  Lots of government departments calling us, taking a somewhat creepy interest in our findings.
[15:55.750 --> 16:01.170]  So, it also inspired other researchers to take a look at broadening these findings.
[16:01.430 --> 16:09.230]  So, this is research by John Penny at the Oxford Internet Institute that came out in March of this year.
[16:10.030 --> 16:11.270]  Looking at Wikipedia.
[16:11.970 --> 16:15.650]  How did that change before and after the prison revelations?
[16:15.790 --> 16:18.850]  And looking at 48 terrorism-related Wikipedia pages.
[16:19.610 --> 16:25.910]  Major Wikipedia pages with a lot of traffic, but things like the Wikipedia page for Heads of Balar.
[16:27.550 --> 16:35.370]  And again, rating the topics independently by Amazon Mechanical Turk raters to provide some objectivity in the process.
[16:35.370 --> 16:43.370]  The findings in the Penny study were a sharp decrease in traffic to Wikipedia pages related to terrorism after the prison revelations.
[16:44.210 --> 16:49.610]  People going much less to find out about terrorism-related traffic.
[16:49.770 --> 16:57.150]  Again, most of the people going to the Wikipedia page for Heads of Balar are not looking to join Heads of Balar, but they're looking to find out about the world around them.
[16:57.370 --> 16:59.550]  And that is the traffic that is being suppressed.
[17:00.170 --> 17:06.090]  No such effect was seen for security-related Wikipedia pages such as the Department of Homeland Security.
[17:06.470 --> 17:11.670]  So, that slants a little bit what people are trying to find out about.
[17:12.470 --> 17:19.890]  And then there was a third study which came out in April, which was completely not based on our research, but still provides some corroborating evidence.
[17:20.450 --> 17:27.270]  Which comes out of the journalism literature from Professor Stoycheffer at Wayne State University.
[17:27.590 --> 17:37.730]  And she ran a lab experiment to examine how people's thinking about surveillance affects people's willingness to express opinions online.
[17:37.730 --> 17:39.550]  Very interesting experiment.
[17:39.710 --> 17:40.370]  I thought it was great.
[17:41.170 --> 17:44.190]  A fake Facebook news post about airstrikes against ISIS.
[17:44.870 --> 17:49.870]  And participants reported how willing they'd be to share their opinions with others on Facebook on the subject.
[17:50.250 --> 17:54.350]  And finally, they were asked about how they thought other Americans felt about the same topic.
[17:54.630 --> 18:04.970]  And for a random subset, they were shown a special notice before they saw the Facebook post, reminding them that the government monitors the online activities of individual citizens.
[18:06.890 --> 18:16.950]  So, here, the findings were that people were likeliest to alter their behaviour after they were reminded about government surveillance.
[18:17.230 --> 18:21.310]  Even if, and especially if, they thought that government surveillance was justified.
[18:21.670 --> 18:26.870]  And they were particularly avoiding sharing opinions that they held, but that they believed were outside the mainstreams.
[18:26.990 --> 18:33.510]  Such as, in this context, maybe airstrikes on ISIS in this story might be a bad idea.
[18:35.010 --> 18:42.630]  And people who said that they had nothing to hide in previous interviews were, in fact, more likely to censor themselves.
[18:42.830 --> 18:44.630]  Which, I think, is illustrative.
[18:44.630 --> 18:46.750]  Because, again, this is a phrase we come across a lot.
[18:47.630 --> 18:53.230]  And maybe there's a certain defensiveness to that phrase that there is, in fact, stuff that people have to hide.
[18:53.230 --> 19:06.570]  This is documentation of a phrase in the literature from Professor Shana Zuboff at Harvard of anticipatory conformity.
[19:06.750 --> 19:07.930]  Somebody knows Zuboff's name.
[19:11.010 --> 19:13.610]  So, people are constraining themselves.
[19:13.910 --> 19:15.370]  This is what these three studies say.
[19:16.210 --> 19:32.310]  They're constraining themselves in what they search for, not only on stuff that they think will get them in trouble with the U.S. government, but also, outside the U.S., in a documented way, stuff that will get them in trouble with their mom if she was looking over their shoulder.
[19:32.490 --> 19:36.590]  And that's kind of, psychologically, how this surveillance is operating.
[19:38.030 --> 19:41.010]  People are constraining what they go out and look for on Wikipedia.
[19:41.730 --> 19:49.810]  People are constraining the opinions that they voice online on controversial topics to do with terrorism and war.
[19:49.810 --> 20:00.350]  So, we find this profoundly interesting and it's something that should feed into the debate over what kinds of mass surveillance should be done.
[20:01.150 --> 20:12.230]  And again, there are going to be a lot of people who look at these findings in the government and surveillance agency world and say, good, we want people to be constraining their opinions.
[20:12.570 --> 20:18.490]  We want people to be suppressing their production of violent extremist content online.
[20:18.490 --> 20:31.330]  We want people to be searching less on anthrax and pipe bomb even if there are legitimate research interests in anthrax and in pipe bombs or if you're trying to find out about the news rather than trying to produce a pipe bomb yourself.
[20:31.930 --> 20:36.910]  So, there is room for that differing viewpoint here.
[20:37.270 --> 20:43.930]  Our perspective would tend to be that chilling effects are bad and that it's evidence that mass surveillance is bad.
[20:43.930 --> 20:47.790]  But we're so black and white about this all, it's disgraceful.
[20:48.550 --> 20:57.910]  So, I want to move on from that to talk about how the social media surveillance systems...
[20:57.910 --> 21:03.110]  A couple of cases where they have led to actual law enforcement involvement.
[21:03.110 --> 21:06.310]  And please don't consider these to be exhaustive.
[21:06.470 --> 21:11.610]  They're simply a couple of many and ones that have happened to happen physically near me.
[21:14.090 --> 21:32.550]  So, one thing to understand is that social media surveillance being conducted by DHS and also by the FBI's Joint Terrorism Task Force, to some extent they're competing systems, tends to ramp up intensely around major public events.
[21:33.230 --> 21:46.770]  And so, in the Boston area, the two biggest public events of the year are the Boston Marathon, which was attacked, and the July 4th celebrations, which were planned to be attacked.
[21:46.770 --> 21:52.430]  So, law enforcement gets kind of antsy around those times.
[21:53.870 --> 22:00.930]  And social media surveillance at the DHS-funded fusion centers and at JTTF ramps up.
[22:01.390 --> 22:05.330]  So, this is one of the things that happened following on from the marathon.
[22:06.430 --> 22:07.350]  There is a guy.
[22:07.550 --> 22:11.430]  He lives in the town next to me, which is Arlington, Massachusetts, or he lived there.
[22:11.430 --> 22:14.350]  His name is Travis Corcoran.
[22:14.590 --> 22:22.950]  And he owned a comic book store and was very into his firearms in the way that many U.S. people are.
[22:24.310 --> 22:29.890]  And he also liked to post vaguely unsettling things on Twitter.
[22:31.390 --> 22:46.770]  Now, he got to here in late June of 2013, so a few months after the marathon attacks, he got to hear about surveillance activists in the Netherlands who were putting party hats on surveillance cameras to mark the birthday of George Orwell.
[22:47.770 --> 22:48.250]  And...
[22:49.350 --> 22:50.510]  It's a cool idea, right?
[22:50.510 --> 22:53.890]  I got to hear about it at the same time and I was like, yeah, you go ahead.
[22:56.210 --> 23:06.990]  But, because I was born in another country and I don't quite get the armaments thing, I didn't do what Travis did, which was on the morning of July 4th to put up this tweet.
[23:07.890 --> 23:14.070]  Putting a party hat on a security camera good, shooting a security camera better, shooting status who in place cameras best.
[23:15.730 --> 23:26.290]  Within the hour, thanks to social media monitoring, there was a 12-man SWAT team round at his house, strongly urging him to allow a search of the premises.
[23:27.090 --> 23:28.630]  He called up his attorney.
[23:28.930 --> 23:33.390]  The attorney said, well, you could say no, but I'm really urging you to say yes here.
[23:33.390 --> 23:35.890]  And so, he let them in.
[23:36.090 --> 23:45.010]  Unfortunately for Travis and for his girlfriend, his girlfriend was cleaning a couple of her legally owned firearms on the bed.
[23:45.130 --> 23:56.330]  And because Massachusetts has relatively strict laws relating to firearms, just like New York State does, they were both charged with felony improper storage of firearms.
[23:57.110 --> 23:59.670]  That is not the interesting part.
[24:00.950 --> 24:15.690]  One interesting element of this is that the charges were dropped in exchange for Travis and his girlfriend setting up the comic book store, setting up their home and moving out of state, so he now lives in New Hampshire because live free or die.
[24:17.270 --> 24:32.750]  The other interesting part is that this case was used by the DHS-funded fusion centre as the key case when talking to legislators about why they are doing good work to thwart terrorism.
[24:34.730 --> 24:40.010]  And we found out about this more or less by accident that they were doing this.
[24:42.830 --> 24:44.890]  So, what does this tell us?
[24:44.950 --> 24:48.250]  This is the best case that they had to bring forward of the good they were doing.
[24:48.470 --> 24:50.790]  And this guy wasn't planning a terrorist attack.
[24:51.050 --> 25:00.750]  He may have been a bloviating guy who likes guns, but ultimately America is full of millions of people who bloviate on Twitter and who like guns, and you cannot arrest them all.
[25:01.970 --> 25:15.430]  But on the other hand, Massachusetts is relatively less full of such people, and maybe it was the violation of a local social norm in that respect that led to more law enforcement attention than he would get otherwise.
[25:16.790 --> 25:22.090]  Next instance of the activities of the local fusion centre.
[25:22.910 --> 25:35.930]  They implemented after the marathon attacks for the Boston Marathon 2014, a geospatial intelligence system that would monitor for threats in real time and map them out.
[25:36.130 --> 25:43.370]  We know this because a fusion centre employee then presented at a geospatial intelligence conference, and this was one of the slides.
[25:43.890 --> 25:50.310]  There were exactly two threats identified at the 2014 Boston Marathon, and both of them are on this slide.
[25:50.310 --> 25:55.910]  One is two Middle Eastern males acting suspiciously asking questions about crowd size.
[25:56.170 --> 26:01.390]  We know nothing further about that because the description is too vague for us to have been able to follow up with them.
[26:01.550 --> 26:05.430]  On the other hand, I do know the journalists at the Bay State Examiner.
[26:06.850 --> 26:07.890]  And they...
[26:07.890 --> 26:25.130]  That's an online publication that, let's say, takes a sceptical attitude towards the surveillance state, and was probing the security arrangements and asking questions such as, do you have a warrant to search any of these bags?
[26:25.770 --> 26:29.750]  And trying to find out whether it really was effective.
[26:31.250 --> 26:35.210]  And they found themselves designated as a security threat.
[26:36.190 --> 26:43.730]  Even though this comes on a slide which is from the fusion centre itself, the fusion centre incidentally completely denies that this ever happened.
[26:45.290 --> 26:59.490]  But it is an illustration that activity on social media, in this case, doing some activity on Twitter, can lead to you being designated as a security threat, even if what you are doing is entirely protected by the First Amendment.
[27:02.630 --> 27:15.250]  Now, there is a growing consensus in government that these kinds of activities are appropriate, and with every terrorist attack they are considered to be more appropriate.
[27:16.370 --> 27:20.990]  The head of GCHQ, the British equivalent of the NSA, came to NSA...
[27:21.530 --> 27:24.110]  Of the NSA, came to MIT and gave a talk.
[27:25.130 --> 27:43.370]  And one of the things he was trying to convey is that he really felt that everyone should be able to agree that what he characterised as the worst behaviour should be driven off major platforms, that there are people who are abusing the Internet to spread violent and extremist ideas,
[27:43.630 --> 27:50.310]  and that it should be a shared goal for customers, industry and government to make that happen.
[27:50.670 --> 27:58.200]  Now, granted, this is Britain, they don't have the First Amendment, but there are similar efforts that are afoot here.
[27:58.790 --> 28:03.610]  It just tends to be done in a slightly more piecemeal and less unitary governmental way.
[28:04.170 --> 28:15.310]  One recent proposal comes from the EU-funded counter-extremism project, which has former Senator Joe Lieberman on its board.
[28:15.310 --> 28:20.230]  Hany Farid is a scholar at Dartmouth University.
[28:21.010 --> 28:42.530]  And this takes its cue from automated DMCA takedown systems, and systems for identifying child pornography on the Internet, and tries to apply it to what they characterise as being violent extremist content, and have that be proactively and in an automated way taken down.
[28:44.170 --> 29:00.290]  And according to press reports, Google and Facebook are both interested in adopting this kind of robust hashing, in order to be able to say to government, look, we are doing something here.
[29:00.290 --> 29:03.170]  We are taking part in the fight against ISIS.
[29:03.430 --> 29:08.590]  We are trying to limit the spread of those materials on the Internet.
[29:09.530 --> 29:17.090]  Which brings us to the awkward question of defining the violent extremist.
[29:17.570 --> 29:28.270]  Here, in this talk, I'm not going to make the argument that ISIS are anything other than mass murdering and raping horrific people or horrific regime.
[29:29.350 --> 29:42.450]  But there's a lot of plasticity to the definition of who is and who is not a violent extremist that comes into play with governmental efforts at social media surveillance.
[29:44.290 --> 29:48.070]  I tried to figure out what an equation for this would look like.
[29:48.330 --> 29:51.490]  And this is more or less what I came up with.
[29:51.490 --> 30:10.250]  that according to police and law enforcement more generally, violent extremists are people who are actually planning violent crimes, which seems entirely fair, plus people who are organising peacefully against the state and its interests, minus violence by the state or people of whom the state approves.
[30:11.490 --> 30:32.390]  And we can see illustratively of this mental attitude, a tweet from the Sheriff of Milwaukee County, who is speculating that Black Lives Matter would join forces with ISIS, which to people who know Black Lives Matter folks seems horrific and absurd.
[30:33.650 --> 30:47.730]  But it is easy, it appears, to define on one side of the line, police and the state and the surveillance agencies and to characterise what they are doing as not violent or extremist.
[30:47.930 --> 31:00.950]  And then on the other side, to take peaceful activists who are sharply disagreeing with the activities of those entities and lumping them together with people who actually are violent and planning violent attacks.
[31:01.310 --> 31:06.990]  So it is something to be exceedingly cautious of as a pattern.
[31:07.550 --> 31:23.470]  And my personal opinion here is that we cannot allow governments to work with social media companies or tech companies to define ahead of time who is and who is not a violent extremist and to suppress speech accordingly.
[31:24.090 --> 31:28.990]  And we have some examples here of how this can start to go wrong.
[31:29.570 --> 31:35.290]  At the top we have DeRay McKesson and Jonetta Olsey, who are two of the leaders of the Black Lives Matter movement.
[31:35.290 --> 31:46.890]  They are already actively under FBI surveillance and have been defined in FBI documents that have been leaked as being violent extremists, even though they have not done anything violent.
[31:48.630 --> 31:57.390]  Leslie Pickering, a few of you may have heard of, he runs an environmental bookstore up in upstate New York.
[31:58.050 --> 32:09.690]  And back in the early 1990s, he acted as a spokesman for the Earth Liberation Front and supported publicly the destruction of property in the course of their activities.
[32:09.950 --> 32:12.450]  But he's never advocated violence against people.
[32:13.690 --> 32:17.610]  And Ross Nielsen, I can almost guarantee none of you will have heard of.
[32:18.830 --> 32:33.170]  But he was identified last week as a violent anarchist for having set up a Facebook page with resources for people who wanted to protest the Republican National Convention in Cleveland.
[32:33.170 --> 32:42.930]  None of which resources advocated violence, but he did have a black and red flag at the top of the Facebook page, and maybe that was enough.
[32:44.950 --> 32:57.630]  So, when we have these unacceptably broad definitions of who is and who is not a violent extremist, and when we go about trying to investigate all of these leads, then we run into deep problems.
[32:57.630 --> 33:06.850]  The FBI actually has a stated policy that they are going to chase up on every lead in the terrorism space.
[33:06.990 --> 33:15.070]  And when you chase up on every lead, then you run into an awkward problem, which some of you will be familiar with, which is the false positives problem.
[33:15.070 --> 33:29.190]  So, the vast majority of people who are posting on any of these topics that are of interest to the government are innocent of any malicious activity.
[33:29.930 --> 33:43.650]  And when you conduct mass surveillance according to search terms on the Internet, or according to where people go on the Internet, the rate of false positives to actual people involved in terrorism is extraordinary.
[33:44.070 --> 33:50.890]  According to various analyses that we've done, sometimes we come up with a figure of 9,000 false leads for every true one.
[33:51.910 --> 34:12.230]  And if you have a policy because you don't want to be seen to be missing out on any terrorism lead, of investigating every terrorism lead, then you run into a situation where you don't know which of these people who is posting stuff that is maybe controversial,
[34:12.230 --> 34:20.630]  like Travis Corcoran's stuff, which of them is going to turn out to be planning an attack and who isn't, and the vast, vast, vast majority of them are not going to be.
[34:21.010 --> 34:24.370]  It's like, have any of you ever stayed in a hospital overnight?
[34:25.190 --> 34:25.670]  Yeah?
[34:25.830 --> 34:26.250]  You have?
[34:26.530 --> 34:31.570]  And there are beepers going off all the time.
[34:32.010 --> 34:35.030]  Every 5 minutes, every 10 minutes, it's hard to get any sleep.
[34:36.670 --> 34:43.470]  And all of these alarms in themselves are valid according to the metrics that the machines are using.
[34:43.470 --> 34:48.010]  But the nurses know that many of them are going to turn out to be false.
[34:48.350 --> 34:52.630]  It's just that they don't want to miss out on anything, and so the result is that nobody gets any peace.
[34:52.990 --> 35:04.110]  Well, in response to the September 11th attacks and the attacks ensuing from it, we are in a situation where nobody is getting the peace from government that they should be able to get.
[35:06.870 --> 35:14.890]  So, I have enthusiastically trashed what the government is doing on social media surveillance.
[35:15.130 --> 35:22.810]  I have enthusiastically trashed their efforts to infiltrate and disrupt peaceful activist groups of various kinds.
[35:22.810 --> 35:33.950]  And so, it would be entirely legitimate to ask me, well, you're so smart, what the hell do you think should be done in response to these kind of attacks?
[35:35.970 --> 35:44.490]  And to be honest, I'm kind of suspicious for the reasons stated that what the government will do could actually be helpful.
[35:45.290 --> 35:52.270]  I tend to be a strong opponent of a militaristic foreign policy as well, and I tend to think that that doesn't make things better.
[35:52.530 --> 36:06.910]  But there are a few suggestions that I have for you and I, for non-governmental people, excusing the agents in the audience, to have things that you and I can do that will help.
[36:07.130 --> 36:17.770]  They won't necessarily prevent attacks as such, but they may decrease this sense that we are heading into an abyss where everyone is being pushed towards the extremist edges.
[36:18.670 --> 36:26.730]  So, one thing to do is to listen to what the president said in a completely different context, which is in the context of campus speech.
[36:27.650 --> 36:32.970]  And he was saying, you don't have to be fearful of somebody spouting bad ideas, just out-argue them.
[36:33.230 --> 36:44.510]  Now, in the context of suppressing violent extremist content online, it can actually be helpful to have revolting ISIS stuff up online and then to have people challenge it.
[36:44.670 --> 37:01.330]  I have seen people take ISIS recruitment and training videos and repurpose them with the theme tune of Yakety Sax, and it's hilarious because they're not really very good at what they're doing, and it does ridicule what they're doing.
[37:01.530 --> 37:19.950]  I have seen people challenge extremist views expressed online in all sorts of helpful ways, and when they do, it is a way of expressing community opposition to this that does not involve the government telling you what you can and cannot say.
[37:20.370 --> 37:28.310]  So, in that sense, I encourage more practical day-to-day solidarity in our everyday interactions.
[37:28.310 --> 37:43.850]  I encourage reaching out across barriers of race and class and gender to listen to people on the other sides of those barriers and to try and find ways to articulate common principles and common thoughts.
[37:44.070 --> 37:48.110]  It doesn't have to be the government mentoring that process.
[37:48.110 --> 38:07.810]  It doesn't have to be them doing what they're doing with countering violent extremist programs in the U.S. and targeting the Muslim community and trying to prevent radicalization by having psychologists report back to the FBI on what young people in those communities are thinking.
[38:07.890 --> 38:12.170]  It can be something much less sinister and much more bottom-up.
[38:12.170 --> 38:18.430]  The problem here is that politicians get asked to do stuff in response to every attack.
[38:19.050 --> 38:32.930]  And what I hear again and again is that politicians do not want to feel like they will be held responsible if there is a major attack and if they can be said to have not done everything possible to thwart it.
[38:33.090 --> 38:37.150]  But the counter example there, frankly, is President George W.
[38:37.310 --> 38:43.650]  Bush, who allowed a fairly major terrorist attack to happen on his watch, and he was rewarded electorally for it.
[38:44.350 --> 39:04.130]  The last recommendation that I have is there is a great deal of talk in the surveillance world about how you need to gather as big a haystack as you can, because if you don't have a big haystack, how can you possibly find the needles of violent extremist content of people planning terrorist attacks?
[39:04.130 --> 39:06.530]  Our perspective is a little different.
[39:06.790 --> 39:13.790]  At Restore the Fourth, we care a lot about the Fourth Amendment, which, for a refresher, for people who don't know it by heart yet, is here.
[39:14.990 --> 39:19.570]  And the perspective of the Fourth Amendment is really that you collect less hay.
[39:19.570 --> 39:43.490]  You only investigate and search and seize the papers and effects and online material of people who you have evidence are involved in actual crimes on an individual basis, not a mass basis, not gathering everything and then flagging what looks suspicious.
[39:44.270 --> 39:55.750]  We believe that this is a law enforcement approach that actually helps law enforcement, because you don't have FBI agents having to chase up enormous numbers of false leads.
[39:56.370 --> 40:00.430]  You may miss stuff if you collect less, but we're missing stuff at the moment.
[40:00.950 --> 40:02.090]  Attacks still happen.
[40:02.330 --> 40:11.830]  Even in France, where they've gone much further towards martial law than we have, attacks are still happening, and it's an unhelpful ratchet to be on.
[40:12.550 --> 40:20.850]  We need a better approach, an approach rooted in the Constitution, where you gather less, but you follow up on it thoroughly.
[40:21.150 --> 40:23.950]  And that is the wisdom of the Fourth Amendment here.
[40:24.570 --> 40:35.810]  So, Restore the Fourth is having a social evening at 8.15pm tonight at the Stoltz Pub, which is just around the corner on West 33rd.
[40:35.990 --> 40:41.450]  We encourage you to join us if you want to learn more about what we're doing around the country at both the national and local level.
[40:42.230 --> 40:45.950]  And at this point, I would welcome questions from the audience.
[40:54.080 --> 40:56.000]  Yes, impressively bearded gentlemen.
[41:04.020 --> 41:06.560]  I have come across the Third Amendment argument.
[41:06.820 --> 41:13.860]  It's an interesting one, though I find it difficult to believe that a court will accept it.
[41:15.040 --> 41:17.260]  But, any solution is worth trying.
[41:17.480 --> 41:20.700]  I feel that the Fourth Amendment is a little more on point for it.
[41:20.860 --> 41:24.960]  But, I'm very happy for people to try whatever strategies they feel will work.
[41:26.900 --> 41:27.460]  Steve.
[41:37.790 --> 41:38.350]  Yay.
[42:04.740 --> 42:11.300]  I think the exact phrase poorly spelled was, F*ck the Boston Bombing, I'm going to go one better.
[42:11.580 --> 42:13.040]  What are I going to do with all you haters?
[42:13.600 --> 42:14.280]  Well, that's just one.
[42:20.540 --> 42:23.880]  The State Police came to his house, a lot of the FBI at the U.S. Martin.
[42:24.960 --> 42:27.240]  He was put in a maximum security sentence.
[42:29.040 --> 42:34.680]  He had the highest bond of anyone in that maximum security prison.
[42:35.440 --> 42:41.680]  Eventually, after, you know, many of the court hearings, the charges would drop any of these funds.
[42:42.180 --> 42:45.060]  But, you know, it's a little ridiculous.
[42:45.480 --> 42:46.280]  It is.
[42:46.440 --> 42:46.900]  I agree.
[42:48.240 --> 42:48.980]  That's a comment.
[42:49.020 --> 42:49.660]  Yes.
[43:36.680 --> 43:38.940]  I think that's a fair comment.
[43:39.120 --> 43:44.240]  And there certainly have been FBI efforts to disrupt white supremacist groups.
[43:46.340 --> 44:02.300]  The difficulty here, when you're analyzing the surveillance state, is that you kind of go with the data that you have out there, and we have more of an understanding of what this unit in DHS does than what the JTTF directly searches for.
[44:02.520 --> 44:11.460]  So, if anyone has access to that information from within FBI, then we would be deeply interested, not that we're advocating anyone doing anything illegal.
[44:11.640 --> 44:11.800]  Yes?
[44:12.380 --> 44:12.760]  Go ahead.
[44:12.760 --> 44:14.100]  The last part.
[44:14.720 --> 44:20.720]  My ponies, if you look up the term brony, that will explain all of your questions.
[44:24.940 --> 44:28.840]  I have daughters who are twins who are eight and a half.
[44:29.020 --> 44:30.920]  I am familiar with the brony-verse.
[44:33.860 --> 44:42.140]  Being kind of a little bit on the inside, on the corporate side, I'm seeing a lot more surveillance taking place in the corporate world and being sold back to the U.S. government.
[44:42.920 --> 44:50.220]  Big data like you would not possibly imagine, and connections know what you really see, are being made on a regular basis.
[44:51.600 --> 44:55.800]  I'm thinking that the only way that you really know this is further obscurification.
[44:56.140 --> 44:57.320]  What do you think about that?
[44:58.120 --> 45:01.180]  I think that that can be a valid strategy.
[45:02.560 --> 45:13.860]  The way that we characterize it at Restore the Fourth is that there are individual things that we can do in terms of obfuscation, in terms of encryption, and these are useful.
[45:13.860 --> 45:18.000]  And we work on legislative advocacy and we work on litigation.
[45:19.460 --> 45:24.340]  None of these strategies disparage any of the others.
[45:24.700 --> 45:30.920]  And self-protection with your data is hugely important when it comes to this.
[45:32.120 --> 45:38.400]  We just have to work on this on whatever angles we can find and that we can develop confidence in.
[45:40.140 --> 45:41.180]  Thank you, first of all.
[45:41.240 --> 45:41.960]  That was a great talk.
[45:43.000 --> 45:55.080]  Are you or any people at Restore the Fourth working on looking at, or are engaged with research on the long-term psychological effects of self-censorship?
[45:55.460 --> 45:58.200]  Or how does that play into the work that you do?
[45:58.200 --> 46:01.920]  I think that would be a fascinating area to look at.
[46:02.160 --> 46:19.620]  I think that, personally, as a researcher, my orientation is more towards empirical techniques, and so I wouldn't feel like I was a natural for doing that kind of longitudinal study.
[46:21.320 --> 46:29.040]  But we do have a 501c3 arm that tries to encourage research into the surveillance state.
[46:29.220 --> 46:31.680]  And so that's something that we can certainly discuss.
[46:31.680 --> 46:34.920]  And my cards are up there if you want to talk about that at greater length.
[46:35.080 --> 46:35.780]  Someone at the back.
[46:37.360 --> 46:42.060]  If I'm somebody who doesn't use social media, but the rest of my family is,
[46:47.040 --> 46:56.820]  I think this is a topic that comes up again and again, particularly when it comes to browsing.
[46:57.140 --> 47:00.460]  Like, if you use Tor, is it labeling you as a suspicious person?
[47:00.460 --> 47:03.060]  If you are not on social media, does it label you?
[47:03.280 --> 47:15.020]  Facebook has shadow profiles of people who are not on Facebook yet, but that it develops in case they ever want to come onto Facebook, which are internal to them.
[47:15.640 --> 47:24.760]  So, it is hard to say exactly what is being inferred from it without being inside the system.
[47:26.420 --> 47:41.280]  It is unfortunately the case that as our public lives become more digitized and more commodified, it becomes less and less possible for us to participate in the world without being surveilled in some way.
[47:41.920 --> 47:45.120]  And there are people who go further to try to avoid this than others.
[47:45.440 --> 47:55.040]  Richard Stallman, who is speaking tomorrow, has gone further than almost anyone in trying to cleanse his profile of anything that is available.
[47:56.280 --> 47:58.580]  I frankly don't go very far.
[47:58.580 --> 48:00.840]  I do have profiles on social media.
[48:01.920 --> 48:06.600]  I do use encrypted communications for some things, but not for others.
[48:08.360 --> 48:17.620]  And my general cross-grained attitude is that I should be able to use these services and have them not be used against me to lock me up.
[48:17.780 --> 48:22.040]  But that is an idealistic perspective, which is why I am in an idealistic line of work.
[48:25.270 --> 48:33.790]  The data that you presented seemed to be mostly oriented on social media postings and searching.
[48:34.180 --> 48:49.120]  And I'm curious if there's been any data on showing effects on one-to-one communication, especially because I would say the greatest revelation from Snowden was in regards to one-to-one communication.
[48:49.120 --> 48:59.310]  I found that, personally, I was restricting what I was talking about with my colleagues one-to-one through digital means.
[48:59.580 --> 49:13.660]  And then once applications like Signal and such arose where I could use those to talk securely, I found myself freeing up again.
[49:14.960 --> 49:24.810]  So, you know, I'm wondering, you know, I'm curious from the room at large, if people bump into me, I'd be curious to hear if other people have similar attitudes.
[49:25.020 --> 49:30.330]  But I'm wondering if there's hard data on that chilling of one-to-one, that effects on one-to-one.
[49:30.710 --> 49:31.980]  It's an excellent question.
[49:32.200 --> 49:52.160]  And when it comes to that sort of one-to-one analysis of infiltration, the best recent study that I'm aware of is a qualitative study out of the UK that deals with the police infiltration and surveillance of activist groups that is being analysed in the Pitchford Inquiry.
[49:53.210 --> 50:12.270]  And it is clear in the context of those activist groups that fear of infiltration and fear of individual conversations being compromised had a significant effect in inhibiting their activities, which were mostly environmental in that context.
[50:12.520 --> 50:15.880]  I can ferret out that study and send it to you.
[50:16.890 --> 50:20.160]  But, so, it does happen.
[50:20.900 --> 50:32.180]  But that kind of analysis is inherently qualitative and is therefore more vulnerable to people saying, well, it's just those people, maybe it doesn't matter so much.
[50:32.290 --> 50:49.200]  The merit of what the Snowden revelations enabled was to be able to analyse the online activities of hundreds of millions of people at a time and to see that there was an effect that was very broad across society that was not limited to people who are activists.
[50:49.500 --> 50:51.520]  And I think that's helpful for people to know.
[50:52.180 --> 50:53.620]  I just wanted to make a comment.
[50:53.830 --> 51:02.600]  I see a lot of optimism in the rebirth of people remembering that the net was a peer-to-peer infrastructure originally.
[51:02.900 --> 51:21.120]  And if we can get away from using these big central data warehouses like Facebook and Google and shift slowly to either a mesh-based network where we all get together and have some sort of free Internet and then that goes on to the public pipes and anonymity is built in and encryption is built in from the start,
[51:21.310 --> 51:33.250]  I think, technologically, we can kind of do an end run around some of these things if we can just wean ourselves off of these big, essentially, data warehouses and go back to more peer-to-peer communication on the net.
[51:33.380 --> 51:35.060]  And IPv6 is a big part of that.
[51:35.200 --> 51:36.600]  And of course, encryption is too.
[51:36.710 --> 51:39.400]  And it's important that we have to keep that encryption in our toolbox.
[51:39.400 --> 51:47.420]  But I see a lot of hope here to kind of get away from... make it very difficult for authorities to trace down exactly who's, you know, saying what.
[51:47.620 --> 51:49.040]  There's a lot of hope technologically.
[51:49.180 --> 51:50.200]  Politically, it's not going to happen.
[51:50.420 --> 51:50.560]  Yeah.
[51:50.860 --> 51:59.770]  I agree that mesh nets are a very interesting and useful potentiality here.
[52:00.920 --> 52:04.940]  And I would encourage people to investigate and adopt them.
[52:05.700 --> 52:10.620]  As with any of these fairly new ideas, the key is usability.
[52:10.620 --> 52:12.980]  And we saw that with encrypted communications.
[52:13.380 --> 52:17.770]  That PGP was a very hard thing for members of the public to use.
[52:17.980 --> 52:19.290]  And Signal is a lot easier.
[52:19.560 --> 52:29.730]  And I've really appreciated the focus on the user experience to make it simple for somebody like me who studied medieval literature at college to be able to access that.
[52:30.180 --> 52:30.660]  Jeff?
[52:30.660 --> 52:53.680]  Do you think that there are any overlaps between studies of effects of surveillance on people for non-political reasons, such as parental surveillance of children or surveillance done by spouse abusers on victims of domestic violence that would be useful for understanding the effects of political surveillance?
[52:53.730 --> 52:58.360]  Or is there something fundamentally different about government involvement?
[52:59.560 --> 53:02.160]  I think that those studies could be useful.
[53:02.310 --> 53:05.500]  And if you have good ones, then I would love for you to send them to me.
[53:06.520 --> 53:20.040]  One thing that we see here is that for a lot of people in these systems, they are thinking of U.S. government surveillance as being analogous to surveillance by a parent.
[53:20.500 --> 53:24.160]  They are thinking of it not necessarily as big brother, but maybe as big mother.
[53:26.080 --> 53:33.440]  And so they are chilling the stuff that is embarrassing that they wouldn't want their mom to see, as well as the political stuff.
[53:33.440 --> 53:40.790]  So that is indicative that that kind of research might be able to offer insights into the political surveillance side.
[53:40.980 --> 53:41.120]  Yes?
[53:41.120 --> 53:43.420]  So I work for the ACLU in California.
[53:44.000 --> 53:44.270]  Awesome.
[53:44.540 --> 53:46.560]  So first of all, this work is tremendously helpful.
[53:46.860 --> 53:47.380]  It's all over.
[53:47.730 --> 53:48.660]  And pushing this over.
[53:48.830 --> 53:53.980]  But one reason I want to point out is that when you talk about these details by the government, increasingly it's not just the U.S. government.
[53:54.200 --> 54:02.810]  It is the state of local authorities who are taking trickle down tools from the government, who are buying off-the-shelf commercial tools, which have some horrendous policies.
[54:02.810 --> 54:05.100]  My favorite example was in Fresno, California.
[54:06.250 --> 54:11.020]  The police officer was using an off-the-shelf tool called media sonar that came with a free package.
[54:11.180 --> 54:20.290]  If you didn't feel like putting in your own watch list or termed in any criminality, they would populate it with black lives matter, hands up, don't shoot, all sorts of other wonderful things.
[54:20.660 --> 54:37.560]  So going forward, it's really important to think about how it's not just big brother up in D.C., but it's all of the other people who are approaching this as a way where they're less sophisticated, often, as I just told you, shows pretty disturbing views of how the society works.
[54:38.250 --> 54:39.960]  I think I was completely right.
[54:40.200 --> 54:54.140]  I had a whole section on fusion centers that I didn't quite get to, but this is something that state and local law enforcement is very keen to get into because it is essentially an infinite spigot of funding.
[54:55.200 --> 55:03.790]  And so we have to be aware of that incentive for them that making their terrorism bust is an important thing.
[55:04.120 --> 55:07.580]  Like with Steve's example from Methuen, Massachusetts of Cameron D'Ambrosio.
[55:07.860 --> 55:12.790]  This is stuff that can make a career in local law enforcement if they make the right bust.
[55:13.060 --> 55:17.770]  So there are powerful impulses towards expanding this kind of surveillance.
[55:20.200 --> 55:23.900]  And that's all we have for questions or time for questions.
[55:24.120 --> 55:29.640]  But if you do have further questions, I'm sure Alex would like to entertain them in the back if that's all right.
[55:29.940 --> 55:30.000]  Yeah.
[55:30.680 --> 55:30.940]  All right.
[55:31.060 --> 55:31.310]  Good man.
[55:31.380 --> 55:31.400]  Amen.
[55:31.700 --> 55:31.710]  Amen.
[55:32.060 --> 55:32.180]  Amen.
[55:32.180 --> 55:32.270]  Amen.