💾 Archived View for tilde.pink › ~ssb22 › ipsos.gmi captured on 2022-04-28 at 17:29:56. Gemini links have been rewritten to link to archived content

-=-=-=-=-=-=-

Ipsos Mori Internet devices

In 2020 we received multiple paper letters saying our address had been randomly selected by a research agency called Ipsos MORI, inviting us to join their “knowledge panel” to answer surveys on public opinion etc. The invitation was non-transferable and required Internet access, but they said if we didn’t have Internet, or if we didn’t think we could use it easily, then they’d give us “a device” to complete their surveys which could also be used for other Internet access.

When talking to them by telephone in December 2020, I was asked the “can you use the Internet easily” question and I gave a slightly-lengthy “it depends” answer—are there any specific browser and OS requirements we had to meet (a consideration which, for the benefit of the presumed non-technical call-centre operator, I simplified to “our computers are getting old and some websites say we need a new one, so I can’t promise every website will fully work” or words to that effect), plus would the site ‘play nicely’ with my accessibility and other settings?—whereupon she decided to sign us up for getting “devices” just in case.

Ipsos arranged for a company called TEK Express to ship us a pair of Galaxy Tab A tablets (made in 2019, weight with cover 685 grammes—both covers’ magnets fell out within 9 months due to light indoor wear-and-tear of the plastic); the tablets initially ran Android 9 and Chrome 87 (later updated), and were locked down by Samsung Knox remote-support software in “Kiosk” mode, to restrict access to anything other than Chrome (without Incognito Mode), Adobe Reader and email. So yes they can be used for “other Internet access” but only as long as this can be done in Chrome on the tablet—they do not provide connectivity to other devices, and you can’t run your own software other than by loading it as Javascript in the browser.

In March 2021 a Chrome configuration update additionally blocked access to 60+ websites, including the videoconferencing services Zoom and Skype, the UK television channels, a dozen subscription film services, Apple, half a dozen ‘livestreaming’ sites, well-known video-sharing sites, half a dozen file-sending and storage sites, Facebook Gaming, and a couple of corporate collaboration tools. This was presumably because each tablet was supplied with a Vodafone data SIM and some users might have been driving up Ipsos’s bills by streaming excessive video over *that* connection—but the block had to apply even on Wi-Fi. Since the block list included sites that were already defunct anyway, I suspect it was an exercise in “what ‘problem’ sites can we think of” rather than being based on actual traffic data.

Another configuration update later in March apparently prevented Chrome from visiting a website on an alternate port number, making it harder to use local test servers running on our Raspberry Pi.
We could however create alternate domains (including wildcard domains) by running an nginx *proxy* on any port, adding this to the Wi-Fi connection settings on the tablet, and adding appropriate connection rules for custom domains in the nginx configuration. That could be used to circumvent the block on alternate port numbers, by assigning custom domains to them instead.
But the YouTube block could not be worked around in this way. While I wouldn’t want to bypass streaming restrictions on *Ipsos’s* data plan, it seemed like ‘collateral damage’ to have them on Wi-Fi too, so I did try setting my Web Adjuster to serve YouTube on a differently-named local domain, but YouTube’s Javascript was too complex for my code to properly adjust (for example their European ‘cookie consent’ logic), and their obfuscation made this too hard to debug.

Knox’s Kiosk mode prevented me from reaching the full Android accessibility settings (indeed it could cause the system to hang if I tried to interact too much with the notification bar, and later versions prevented me from reading the notifications at all, although the status bar still told me that notifications were there to be read somehow); only brightness and limited-range font size options could be set, and when I phoned Support asking how to get out of Kiosk to enable dark mode, they “remoted in” (asking only for the 5-digit “asset number” stuck to the case, so accidental connections to the wrong tablet probably could happen—it didn’t prompt me to allow it; it did show notifications at the start and end of the support session but not during it)—and turned down the brightness control, saying that’s the best we can do (no dark mode). But Android’s “hold both volume keys” shortcut for toggling Voice Assistant *did* work (after being enabled the first time and then repeated to actually switch it on), although you couldn’t change its voice even though the Javascript Web Speech API showed at least three voices available. Some application-level accessibility settings were still available: the Email application’s menu - cogwheel - “Make Email app dark” was allowed but doesn’t apply to the email body, and Chrome’s dark theme (plus, if you’re adventurous, enable-force-dark under chrome://flags) was allowed, although it wasn’t possible to add bookmarks or change the home page.

The Kiosk mode also prevented us from interacting with the software update process. If we turned the tablets fully off when not in use, updates tended not to happen at all (we’d be constantly notified they’re pending but were unable to interact with these notifications) so we had to leave them switched on to standby—but this caused the tablets to periodically play *sounds* when restarting after an update and at other seemingly-random times, and Kiosk prevented us from accessing any setting to disable these sounds. This was not a major problem until one of us accidentally left a tablet in the bedroom overnight and it played its sound at *five o’clock in the morning* on a day when we really needed to have slept well. Do not leave these tablets anywhere where you’d be disturbed by occasional loud beeping sounds at any time of day or night.

Actual completion of their surveys wasn’t too difficult, although they have asked some multiple-choice questions to which my natural reaction would be “it depends what you mean by this word and/or it depends on the exact circumstances—I could write a paragraph about it” but that wasn’t an option so I had to go for “not sure” or “don’t know” which Ipsos apparently tries to discourage by displaying in grey text reminiscent of a disabled form control (perhaps to balance out the temptation to skip time-consuming questions?); I suppose these surveys are not focus groups. Some later surveys did allow free text entry (e.g. what do I think the issues with battery-powered aircraft might be); typing large amounts of text on the tablet would of course have been difficult even with the supplied magnetic pen, but if necessary I could arrange for a paragraph to be temporarily placed on a private website so I can copy and paste it into the survey, and I was also able to make English voice-recognition work on the tablet via a microphone button above the keyboard, although the accuracy of this is variable. We don’t remember being given *all* the UK surveys mentioned on Ipsos’s “News and Polls” web pages, so it seems they limit participant load by not enrolling everybody in every survey.

Ipsos survey “points” could be converted into shopping vouchers for a supermarket of your choice, although “your choice” in this instance is limited to Tesco, Sainsbury’s, M&S, Argos or Amazon online. We were able to redeem discounts at both Tesco and Sainsbury’s by asking a cashier to scan a barcode from a mobile screen, but in the case of Sainsbury’s their cashier’s machine refused to accept the voucher on the first day but accepted it on our following trip. We did not try the vouchers on either shop’s self-checkout machines (we thought this experiment would be likely to delay the queue), and there was conflicting information about whether or not the discounts were permitted for use online so we didn’t try that either. Balance enquiry was by refreshing the voucher page for Sainsbury’s and by telephone for Tesco (0345 number provided) which we confirmed worked: if your shopping bill is less than the value provided, then the voucher is automatically ‘part-spent’ and some balance remains on it. At time of issue the expiry dates were set at 10 years for Amazon, 5 years for Tesco and just over 1 year for Sainsbury’s.

We could not find any facility to access old vouchers via knowledge-panel login: if the email is accidentally deleted then the voucher is lost. But it might not have been deleted: the email application on the tablet came with the account’s “sync period” set to 2 weeks (meaning it would not display any emails older than 2 weeks), so the first thing to try is to set this to “all time” to show older messages.

Legal

All material © Silas S. Brown unless otherwise stated. Adobe is a trademark of Adobe Systems Incorporated in the United States and/or other countries. Android is a trademark of Google LLC. Apple is a trademark of Apple Inc. Facebook is a trademark of Facebook, Inc. Javascript is a trademark of Oracle Corporation in the US. Samsung is a registered trademark of Samsung. Skype is a trademark of Microsoft in the US (but not in Europe because it was too similar to Sky). Tesco is a trademark of Tesco Stores Limited. Vodafone is a trademark of Vodafone Group Plc. Wi-Fi is a trademark of the Wi-Fi Alliance. YouTube is a trademark of Google Inc. Zoom is a trademark of Zoom Video Communications, Inc. Any other trademarks I mentioned without realising are trademarks of their respective holders.