|
Spangly A posted:Goes to boarding school, engages in weird rituals and initiations, lives in a world that behaves entirerly without the knowledge or consent of the wider population Has an opponent who's a pasty-faced blood purist? Tries to get as far away as possible from his upper-middle-class relatives?
|
#
?
May 13, 2016 16:16
|
|
|
|
| # ? Apr 23, 2024 22:14 |
|
|
FAUXTON posted:It's unfortunate so many people in the US have similar awareness - hands off my medicare and the like. Sure, that's probably largely an effect of "rugged individualist" marketing which downplays or even omits the fact that someone is receiving publicly funded benefits. People pay for their own medicare / social security with their own taxes (extremely regressive payroll taxes actually, so the poor pay for their own poo poo even harder), its not a freebie program like say welfare or euro type uhc. It makes sense people see it as their own money because it actually is, perhaps you are thinking of medicaid?
|
|
#
?
May 13, 2016 16:31
|
|
|
TROIKA CURES GREEK posted:People pay for their own medicare / social security with their own taxes (extremely regressive payroll taxes actually, so the poor pay for their own poo poo even harder), its not a freebie program like say welfare or euro type uhc. It makes sense people see it as their own money because it actually is, perhaps you are thinking of medicaid? Those benefits are drawn from present revenues, and the current beneficiaries paid for their parents' and grandparent's benefits when they were in the workforce. It isn't their own money and it never has been, perhaps you are thinking of IRAs and HSAs?
|
|
#
?
May 13, 2016 16:38
|
|
|
-
crowoutofcontext fucked around with this message at 21:37 on May 14, 2016 |
|
#
?
May 13, 2016 18:29
|
|
|
Lotta problems with that visualization- in addition to the notability problems coming from a wiki sample, and other wiki errors (the tan "murd" on the right stands for "murderer"), it appears to assume, or at least can't represent exceptions to, exclusivity. So, for example, how does it handle a politician that is also a businessman?
Discendo Vox fucked around with this message at 18:48 on May 13, 2016 |
|
#
?
May 13, 2016 18:40
|
|
|
Yeah that seems pretty useless. The wiki data he used has 194 rows. The Officers table from the panama papers database has around 345,000, though some of these aren't people. If you want to go down that route, you'd probably start with a random sample of 3-400 and go from there, rather than starting from the most notable people identified.
|
|
#
?
May 13, 2016 19:03
|
|
|
I, for one, am a little taken aback at the size of "Journalist" on there.
|
|
#
?
May 14, 2016 00:30
|
|
|
Captain Bravo posted:I, for one, am a little taken aback at the size of "Journalist" on there. I could see the secrecy for certain types of investigative journalism. Unfortunately, the best places to make obfuscatory shell corporations are also tax havens so everyone is going to think you're just routing your blood money through the Caymans to gently caress over poor people regardless of whether there's been any money moved within the past decade.
|
|
#
?
May 14, 2016 03:51
|
|
|
Captain Bravo posted:I, for one, am a little taken aback at the size of "Journalist" on there. Most likely explanation is that it's people who own newspapers. Again, though, although the graph is appealing to read meaning into, it's basically a cipher.
|
|
#
?
May 14, 2016 04:26
|
|
|
FAUXTON posted:I could see the secrecy for certain types of investigative journalism. Unfortunately, the best places to make obfuscatory shell corporations are also tax havens so everyone is going to think you're just routing your blood money through the Caymans to gently caress over poor people regardless of whether there's been any money moved within the past decade. I've known investigative journalists, and very, very few pull in enough money to even dream of a shell corporation. I'm thinking the more likely explanation is what Discendo Vox said.
|
|
#
?
May 14, 2016 04:30
|
|
|
crowoutofcontext posted:This is based on tags from a wikidata file on the Pananama Papers, so isn't the truest breakdown of occupations, but is a start to such a complex map: 
|
|
#
?
May 14, 2016 05:02
|
|
|
Huh. That interface is kinda fun to play around with. I can't actually find a full occupation name that starts with "murd" - no idea where that circle comes from. If you go to the Wikidata Query Service and enter the following as your query:quote:SELECT ?person ?occupationLabel WHERE { You'll get a table with a link to the person's entry in Wikidata and their occupation. The four journalists are Uri Blau, Leyla Aliyeva, Sigmundur Gunnlaugsson, and Mario Llosa. ...occupation may not be the most accurate field ever. Edit: Missed an ending curly brace. moebius2778 fucked around with this message at 09:18 on May 15, 2016 |
|
#
?
May 14, 2016 05:21
|
|
|
Cripes. Look, the chart isn't accurate. Please stop using it. Crowoutofcontext, it was a good idea, but would you consider deleting it? I've sent it to some folks whoa acutally deal in social network analysis- one of them is running Sergey Roldugin's "neighborhood" of entities right now.moebius2778 posted:Huh. That interface is kinda fun to play around with. I can't actually find a full occupation name that starts with "murd" - no idea where that circle comes from. If you go to the Wikidata Query Service and enter the following as your query: It was El Chapo. Someone edited his wikipedia article so that his occupation was "murderer". Wiki scraping is not a good way to develop datasets.
|
|
#
?
May 14, 2016 05:29
|
|
|
If you check slightly to the south of "Lawyer" there also seems to be significant corruption among V6s & V4s
|
|
#
?
May 14, 2016 05:54
|
|
|
Yeah, that was very misleading, even if it didn't intend to be.
|
|
#
?
May 14, 2016 06:01
|
|
|
Discendo Vox posted:It was El Chapo. Someone edited his wikipedia article so that his occupation was "murderer". Wiki scraping is not a good way to develop datasets. I was rather hoping it stood for murd-och.
|
|
#
?
May 14, 2016 14:44
|
|
|
Judo?
|
|
#
?
May 14, 2016 16:27
|
|
|
The Larch posted:Judo? Maybe Hong Kong martial arts actors? I love how this chart confuses more than it enlightens.
|
|
#
?
May 14, 2016 16:36
|
|
|
Snapchat A Titty posted:I love how this chart confuses more than it enlightens. It seems that it was someones pet-project on a blog, I knew it'd be a bit problematic but it seems to have no redeeming features. Took it down. Though this is a problem with these massive data-dumps, they demand to be simplified into a simpler narrative if they are going to create a public reaction, which is why the blogger embarked on the confusing info-graph project. Though It might be more interesting (and useful) to discuss the numerous branch cases instead of intellectually choke one self by trying to ingest this information in one go.
|
|
#
?
May 14, 2016 21:59
|
|
|
crowoutofcontext posted:It seems that it was someones pet-project on a blog, I knew it'd be a bit problematic but it seems to have no redeeming features. Took it down. Yeah. It was in the news for a cycle or two in Denmark that Tax are gonna look into "65 subjects" which is what you get if you don't type anything and choose Denmark. 65 or 67 or whatever. Just a set of characters that are already associated with Denmark. No deep look, no extra checks. Denmark literally paid out what amounts to a billion dollars on fraudulent tax claims. http://uk.reuters.com/article/uk-denmark-tax-fraud-idUKKCN0QV0SI20150826 I'm so glad they've gutted Tax employment. Obviously we don't need people there anymore 
|
|
#
?
May 14, 2016 22:20
|
|
|
Small update from the SNA people. Part of the problem with the dataset is that in addition to being enormous (you can't open some of it in excel), the whole thing needs to be cleaned. In context that means that right now (for example) misspellings, honorifics, and other variations have created a ton of cases where a person or company is showing up as two separate people in the database, each with a separate set of associates. The date information is also really inconsistently structured, making it hard to do reliable longitudinal analysis. Someone's probably going to need a grant and some staff to clean the whole thing.
|
|
#
?
May 15, 2016 00:13
|
|
|
The other challenge with any sort of SNA is that there are a lot of people who are connected to others, but it's through weirdly structured intermediaries rather than because of an actual connection. Like this guy is connected to dozens of random people if you follow through all the companies he's connected to. But it's because he works for some business registration company. I suspect there's hundreds of cases like this, so a lot of the connections you'll find are rather tenuous.
|
|
#
?
May 15, 2016 00:22
|
|
|
Xandu posted:The other challenge with any sort of SNA is that there are a lot of people who are connected to others, but it's through weirdly structured intermediaries rather than because of an actual connection. Like this guy is connected to dozens of random people if you follow through all the companies he's connected to. But it's because he works for some business registration company. I suspect there's hundreds of cases like this, so a lot of the connections you'll find are rather tenuous. Yeah, one of the people involved is trying to partition it, then project it into a network of entities or a network of officers/intermediaries. It's rapidly getting beyond my SNA knowledge, though- I was planning to use SNA for my diss, but what I wanted to do requires methods that haven't been developed yet, plus other requirements of the degree got in the way. In the case of people like the one you mention, you'd need to construct another sample out of the data using the companies these officers work for. That wouldn't need to be a network, but it would let you know what companies are selling offshore systems--and it would give you exclusion criteria for their employees in the main dataset, if such companies aren't what you're interested in. Discendo Vox fucked around with this message at 22:54 on May 18, 2016 |
|
#
?
May 15, 2016 00:28
|
|
|
i wish i was kidding when i say: 1%ers up against the wall. sure, some are "innocent" but i dont mind it really
|
|
#
?
May 15, 2016 01:11
|
|
|
The Larch posted:Judo? UFC fighters? Ronda Rousey comes to mind.
|
|
#
?
May 15, 2016 06:45
|
|
|
Snapchat A Titty posted:i wish i was kidding when i say: 1%ers up against the wall. nah none spared no remnant saved
|
|
#
?
May 15, 2016 08:55
|
|
|
kill the rich
|
|
#
?
May 15, 2016 10:32
|
|
|
Folks have isolated the networks for Russia, Greece and Cyprus. They're working on creating networks that isolate and remove address and entity nodes as needed for analysis. It looks like data cleaning is going to be piecemeal, arranged by country at first. I'd send a link to the network output image, but it's not very useful yet (it's unlabeled analysis software output), it's semi-proprietary to the people who created it and it's still so large that it takes about 15 minutes to load. Heckwithit, here's a screenshot. Please don't spread it around.  See the large blue cluster in the lower right? That's Russian, and there are two nodes (I can't tell what they represent) at the center of the whole thing. The other big russian cluster at the upper left has a single node at the center. All the seemingly unattached nodes around the outside of the clusters look that way because this network only includes the three countries. Those nodes will have ties to other countries, as will the ones in the clusters- the international ties aren't visible here. Again, this will only start being really useful when we can filter for different relationships in the network.
|
|
#
?
May 15, 2016 22:03
|
|
|
What's the next step with that stuff as far as analysis goes? I've always thought those network graphs looked fantastic, but what does it tell you?
|
|
#
?
May 16, 2016 05:02
|
|
|
Xandu posted:What's the next step with that stuff as far as analysis goes? I've always thought those network graphs looked fantastic, but what does it tell you? Again, caveat that this is rapidly going beyond my SNA knowledge, so I can't explain the procedures in detail. They're going to subdivide the network by what each node represents: corporations, officers, addresses, etc. Once they've pared the data down to a point where it's subdivided into sets that can be read by humans, people are going to have to clean the data to identify where, for example, "Mr. Xandu" and "Xandu" and "Xandu, Esq" and "Xnadu" are erroneously being treated as separate people, and to distinguish "Xandu" from "Xandu"- unless that's also you, of course. Once the cleaning is done for the subset of interest (let's say all entities with officers in Russia) they can do a number of different data transformations to focus on particular questions. You have to construct different networks to test different hypotheses. The data cleaning, and assembling an overarching network, will give them an opportunity to eyeball some general configurations and numbers to have an idea of what questions, what tests, would be good to ask/run. For example, let's talk about those clusters. They actually have a common format: A. one or two central core nodes, B. a "layer" of nodes that make up the cluster, each of which are connected to the core node and also connected to C. one other outer node. These outer nodes are arranged in a sort of ring around the periphery of the cluster. My suspicion is that the core node at A is a company that's selling offshore shell entities, the layer at B is the companies they've sold, and that C is the clients, who are listed as the officer and sole shareholder of the company in question. This is just a guess until we also subdivide all the nodes to identify what they are. Once we do that, though we can use other outside information to ask other information about the network- for example, what proportion of people with offshore entities sold from Company X are located in, e.g., Mexico, and how many of them are friends with Guzman? When did that practice start, and when did it end? More interestingly, how many countries are such associated accounts passed through, and are there commonalities in the route? Not all of this process requires SNA, but it can also be used to develop a statistical model that can tell you the what money laundering in a network of shell corporations looks like. That model could be applied more generally by regulators. More fruitfully (and closer to what I've seen SNA used for), it can be used to identify enforcement targets that will disrupt illegal activity for a longer period in more complicated networks(SNA researchers have used the same analysis to decide which members of terrorist networks to target for drone strikes). The graph I posted, because it's cut off so much info, makes the overarching network seem a lot simpler than it is-in practice really complicated, SNA-specific methods are needed. Discendo Vox fucked around with this message at 08:06 on May 16, 2016 |
|
#
?
May 16, 2016 07:52
|
|
|
Discendo Vox posted:Heckwithit, here's a screenshot. Please don't spread it around. SA isn't exactly a private discussion forum.  What's the difference between the blue and gray/greenish nodes?
|
|
#
?
May 16, 2016 22:24
|
|
|
Fuschia tude posted:SA isn't exactly a private discussion forum. I know it's not private, but neither's the data. Just don't go posting this unlabeled graph to Facebook. The color is country; blue is Russia, green is, um, Cyprus I think.
|
|
#
?
May 17, 2016 00:26
|
|
|
Discendo Vox posted:Folks have isolated the networks for Russia, Greece and Cyprus. They're working on creating networks that isolate and remove address and entity nodes as needed for analysis. It looks like data cleaning is going to be piecemeal, arranged by country at first. I'd send a link to the network output image, but it's not very useful yet (it's unlabeled analysis software output), it's semi-proprietary to the people who created it and it's still so large that it takes about 15 minutes to load. Are these people in contact with the OCCRP? If not, I can get them in touch.
|
|
#
?
May 18, 2016 19:07
|
|
|
snip.
Discendo Vox fucked around with this message at 22:46 on May 18, 2016 |
|
#
?
May 18, 2016 19:47
|
|
|
|
| # ? Apr 23, 2024 22:14 |
|
|
Discendo Vox posted:It does not appear so. What I have is access to a general SNA professional organization mailing list. If you'd like, I can PM you details and we can discuss whether it would be appropriate-there may be some political issues to consider. I'm reaching out to folks more familiar with the SNA research field than I to find out who best to connect you with. Sure, pop me a DM, I know the OCCRP team well, just met with them last week to discuss things.
|
|
#
?
May 18, 2016 22:10
|
|