Evaluating People-Search Site Removal Services

August 9, 2024 1:15 PM

New Consumer Reports study on getting your address, phone number, etc. removed from sites like Spokeo: "Data Defense: Evaluating People-Search Site Removal Services". "a group of companies offers to remove people’s personal data from people-search sites for fees ranging from $19.99 per year to more than $1,000 per year..." They tested 7 removal services to try to delete participant info from 13 people-search sites. While the study's authors "do not consider the results statistically significant or nationally representative", they found that "As a whole, people-search removal services are largely ineffective." and "Manual opt-outs were more effective than people-search removal services but were also far from perfect." (19-page PDF including Methodology and Limitations sections.)
posted by brainwane (18 comments total)

What in the world does “the results are not statistically significant” mean??
posted by MisantropicPainforest at 1:45 PM on August 9

MisantropicPainforest, I think that they are just pointing out that testing 7 people-search site removal services and finding them ineffective does not prove (though it might suggest!) that all such services are ineffective.

(I originally inadvertently compressed "people-search site removal" to "people-removal" in my comment. Such services also seem, in the instances that come to light, to be largely ineffective.)
posted by It is regrettable that at 1:50 PM on August 9 [2 favorites]

summary table for the lazy - pretty much sums it up, just remove manually and don't bother paying
posted by Rhomboid at 2:39 PM on August 9 [4 favorites]

I have manually opted out of all the services, it’s really easy and doesn’t take long. I didn’t realize there were services that would charge to do it for you but I’m not surprised.
posted by Slinga at 3:04 PM on August 9

all of the services? How can you know? If an aggregator is storing data offshore, there's nothing you can do about it
posted by scruss at 3:19 PM on August 9 [3 favorites]

I assumed that meant all of the services in that table. Or all of the big ones.

In reality you will probably never know all the places holding a data portfolio on you, especially if you are in the US. But this was specifically the weird category of “people search sites.”
posted by teece303 at 7:41 PM on August 9

MisanthropicPainforest: What in the world does “the results are not statistically significant” mean??

I've grown up mathsy in a science household, so assume this shorthand has become well known everywhere.

There's a background assumption to all scientific tests: this action has consequences that you can distinguish from doing nothing (or doing the best practice, say if you're testing whether a new medicine is a better treatment).

We say "any differences are not statistically significant" when what we did had no impact. There's a bunch of work put in to control the setting where the test takes places so that the only differences are doing the thing and not doing the thing, then we gather at least 20, ideally over 100, repeats of the test so that averaging statistics show results that are generally like the whole population.

So, here "not statistically significant" says that they didn't have enough test data to generalise to the whole population. It might look ineffective, but that might be something unusual among their volunteers to use these services with real-person lives and data.
posted by k3ninho at 11:07 PM on August 9 [2 favorites]

I’m a data scientist. It’s just that considering a qualitative study as ‘not statistically significant’ or ‘statistically significant’ makes no sense, at all. It’s a massive category error.
posted by MisantropicPainforest at 7:21 AM on August 10 [1 favorite]

It's hilarious to me that each and every one of these services has an opt out feature, and all of them require you to enter your email address in order to opt out. It's like a carnival ride you can't get off.
posted by blendor at 7:30 AM on August 10

I have always assumed the opt out features were their verification feature. If you opt out, they know they are correct about you. Or at least correct-ish.

The public facing stuff then stops showing up, but that never really mattered to them any way. It’s what they do with your data behind the scenes that matters. And they now have a little vote of confidence from you personally.

Maybe I’m being too cynical there.
posted by teece303 at 8:46 AM on August 10 [3 favorites]

Can’t we just make it illegal to collect and store personal information like this? Yes, I realize that would destroy the Internet economy if you include social media, targeted ads, data brokers, etc. but I think we should rip the bandaid off.
posted by caviar2d2 at 9:53 AM on August 10 [5 favorites]

Maybe I’m being too cynical there.
Nope, just the right amount of cynical. I went ahead with the opt outs, if only because the hoops they made me jump through to do so were so Byzantine it made me feel like they didn't want me to, so there's that.

I also clicked the Google ad for every one of them just to cost them a little bit of AdWords money, because fuck them.
posted by blendor at 9:57 AM on August 10 [1 favorite]

I think in this case, although as described above it's a massive category error, not statistically significant it intended to mean that the effectiveness of the removal services did not differ from one another.
posted by bluesky43 at 12:46 PM on August 10

Shower thought: there's no meaningful need or obligation to be truthful with data collectors and brokers (at least those who aren't mostly already regulated, like educational, financial or medical enterprises). The only real barrier is issues with account verification/recovery. I wonder if an identity fuzzing service would be effective.
posted by snuffleupagus at 12:48 PM on August 10

Can’t we just make it illegal to collect and store personal information like this? Yes, I realize that would destroy the Internet economy if you include social media, targeted ads, data brokers, etc. but I think we should rip the bandaid off.

Illinois just significantly weakened its bio-metric privacy laws at the behest of the big internet companies so the answer is going to be no. This despite there not being a single settlement that even approached to legal penalty amounts for violations and the settlements themselves were getting smaller and smaller with each case. My last settlement was something like $35 from Instagram despite the law specifying penalties of $1500. With the newly watered down law the payments will likely not even you get a small coffee.
posted by srboisvert at 6:11 AM on August 11 [2 favorites]

Can’t we just make it illegal to collect and store personal information like this?

Realistically, no, this can’t be done in the U.S. Meta, Alphabet et al would sue the government, and with the Supreme Court’s recent week of terror, any such law would probably just be declared unconstitutional by the Supreme Court.
posted by teece303 at 11:26 AM on August 11

The first sentence of the Limitations section of the report (page 9 of the PDF):

This study included a relatively small sample size and participants were selected from among volunteers in the non-random process described above, so we do not consider the results statistically significant or nationally representative.

As the Methodology section (pages 7 and 8) explains, the test used 32 test subjects:

Our final group of 32 participants consisted of four groups: eight New York State residents who own their homes, eight New York State residents who rent their homes, eight California residents who own their homes, and eight California residents who rent.....

With each group of eight participants, seven participants were randomly assigned one of the people-search removal services; for the eighth, we manually opted out of the 13 people-search sites, to set a baseline for comparison. This means that we evaluated each of the people-search removal services with four participants, one from each of the four groups.....

There are a bunch of limitations here -- for instance, all of the volunteers previously "had indicated an interest in data privacy-related projects" which might skew the sample, and "We also eliminated volunteers who had not lived at their current address for at least two years, and those who had highly common surnames, which would make it relatively difficult to isolate their profiles on people-search sites."

Also, btw, in the Findings: "EasyOptOuts and Optery performed the best of the services we evaluated." EasyOptOuts cost USD$20/year (the least expensive of the seven services they tested). Optery, at the tier they tested: $249/year, the most expensive option. So, the cheapest and the priciest options. Weird!
posted by brainwane at 10:01 AM on August 12

BTW, I've corresponded with Yael Grauer, one of the authors, who has been happy to answer my emailed questions.
posted by brainwane at 8:21 AM on August 20