Google's new Chrome extension enables flagging of suspicious sites

[C_Zalcman]

Google today launched a new Chrome version and a new extension Suspicious Site Reporter that allows anyone to flag suspicious sites.
What is interesting is that Google mark as suspicious any site that isn’t a top 5,000 site.

Screenshot from Chromium Blog:
[lh5.googleusercontent.com ]

This is not misleading people?

[Wilburforce]

Well, it certainly seems to have a few questionable aspects, but having had a look at it I'm not sure your statement that "Google mark as suspicious any site that isn’t a top 5,000 site" exactly reflects what it does.

The tool gives a colour-graded "suspicion" flag, with sites that have no grounds for suspicion getting a green flag. those with a few minor grounds amber, while those with many or serious grounds get a red one. Amber and red flags also show a count of possible grounds for suspicion as Google sees them. You may suspect a site for reasons Google hasn't included in the tool.

Most sites I looked at from a couple of reasonably irrelevant searches get an amber flag with a count of 2 (two grounds for possible suspcion):

- Site not in top 5k sites
- Haven't visited site in the last 3 months

It would clearly be possible for me not to have visited a site in the top 5,000 in the last 3 months (althought I haven't yet found one), and my presumption is that a site fitting those conditions would get an amber flag and a count of 1 (Haven't visited site in the last 3 months). If I am correct (please correct me if there is evidence to the contrary), it isn't passing all sites in the top 5,000 and failing all others: it is simply giving a "clean bill of health" to any site with no potential grounds for suspicion at all. Not being in the top 5,000 isn't a "fail", it is advisory, in the same way that not having visited a site is advisory (as it also advises us in the clearest possible way that if you use Chrome Google keeps a record of every site you visit).

I'm not sure what the tool is intended to achieve, how widespread the take-up will be (it isn't added by default, and finding and installing it requires proactive effort by users), or how useful it will be to Google or anyone else. It seems fraught with potential problems, as it is easy to envisage how zealots might use it to report nearly everyone, bad boys might use it to deluge Google with false reports about competitors, and people who are being scammed and defrauded might not have installed it, or might not use it even if they have,

I only use Chrome for browser-compatibility checking anyway, so it isn't much use to me, and having looked at it I will now disable or remove it.

[Robert Charlton]

C_Zalcman, thanks for spotting this... and Wilburforce, thanks for your thoughtful comments. Did you get a sense of what is meant here by "a top 5,000 site"?

Your comment "reasonably irrelevant" suggests you're thinking it might be related to rankings for a specific query, which, if so, triggers off all kinds of thoughts about a "new" Google metric... perhaps tied to topic-related PageRank, about which there's been much speculation, but, if I remember correctly, no official confirmation.

Perhaps a quality related metric, not tied to relevance?

Or, conceivably, it might be a traffic-related metric (like Alexa), though I can't imagine Google giving such a measurement much credibility.

[iamlost]

This looks like another Google getting users to train their ML algos.

They apparently checked the top (by some metric) 5000 sites and are confident they are not deceptive/suspicious. Rather than take on the remaining 100 million or so sites they will off load the exercise. Sort of as they use reCAPTCHA as a human check on the algos in question.

What will be interesting is how soon which sites one visits become 'unmarked'. And, harder to determine, how long it takes a site to become universally 'clear' or perhaps it's truly personalised so sites only 'clear 'change for frequently (quarterly) individuals. Although that could have legal repercussions so I'd think that last scenario unlikely.

Me thinks I see another masked headless Chrome task...

[Wilburforce]

@Robert Charlton

When I had a look at it earlier I deliberately tried searches that probably wouldn't bring up "top 5,000" sites, to see whether that warning applied universally. By "reasonably irrelevant" I just meant things like - actual examples - "garden furniture" and "dairy products", which had no direct bearing on my own interests or any specific sectors or themes I see here.

I also picked things that I thought were unlikely to have "top 5,000" sites in the results to confirm that the "warning" applies to most mundane product, service and information sites. As far as I can see, it applies universally, even to PPC advertisers.

However, on thinking about it further I thought I should try something that would be more likely to bring up "top 5,000" results. Initially I tried "world news", in which (obviously geolocation-biased) the top result was BBC world news. This didn't have the "Site not in top 5k sites" warning, so is clearly one of the top 5,000 sites in whatever metric(s) Google uses. However, it still got an amber flag and a score of 1 ("Haven't visited site in the last 3 months") so my earlier assumption was correct.

I navigated away, then back again to see if the "haven't visited..." warning had cleared, but it hadn't. Closing and reopening Chrome didn't clear it either (it is still there after a couple of hours), which is in itself interesting. Possibly my personal security and privacy preferences in Chrome are being respected (all credit to Google if they are), leaving Google without any information to tie my browser identity to the previous visit. Possibly, however, there is some time-lag or other factor involved.

It is certainly interesting, but I really don't see where they are heading with it. I can't think they will get enough reliable and unbiased human input to hang a coat on, let alone a worldwide "trust" policy. Iamlost's question is interesting too: apart from the 3 months warning (or other changes in status, like entering the top 5,000), will any site ever be "unmarked", and what about going in the other direction: is there a "red-flag" marker that will come up saying "multiple reports of suspicion"?

[Dimitri]

By default, the status of a site should be "neutral", and not "suspicious" until legitimate reports are made, or if Google(bot) detects phishy behaviour.

one is considered innocent unless proven guilty

The flag should be "gray" (neutral) for unreviewed sites, and orange only when something has been reported or detected. Being in the top 5K sites is not a guarantee that the site (or page) is not involved in suspicious activities (I am sure some torrent or file sharing sites are in the top 5K for example, or all kind of sites spreading conspiracy theories...).

Also, is the "report" function to report a site is suspect, or to report it is not ?

[lucy24]

This sounded awfully familiar and then I remembered that a while back I installed the Firefox “Web of Trust” extension ... and almost immediately deactivated it because it’s just more bother. (“Is WebmasterWorld safe? 4.7 out of 5 stars say yes.” By weird coincidence, I see the same rating for google dot com.)

[Marshall]

Another attempt by G to control the internet and push smaller sites from which they do not benefit financially, out of existence in search results, IMHO. "Do no harm" has gone the way of the Dodo.

[Selen]

No worries, all p()rn sites are safe.

[lucy24]

all p()rn sites are safe

... while, in FF's case, they reserve judgment (icon is grey rather than green) on the World Register of Marine Species. Ooh, yeah, that’s definitely a shady* venture. Can never be too safe.

* Obvious pun suppressed.

[tangor]

Mob rule is always a worry ...

Meanwhile, which 5,000 websites are okay? Stating that means all the rest are NOT OKAY. I suspect there's been a few backroom deals made. :)

</tinfoil, hat-shaped perhaps>
</humor if not obvious>

[mcneely]

So Firefox rolled this function out too once upon a time -- Call me a skeptic if you will, but I don't think the average end user is going to even pay attention to this ... They blow through the net on these browsers with absolute abandon and without a care in the world -- And given Google Chrome for the spyware that it is, I'm inclined to imagine this ... /cough cough .. "new function" .. /cough cough ... to be just one more addition to, #1 continuing to log end user surfing habits (keylogging), and #2 to check sites from "not" Google servers (save bandwidth)

The internet is all about hype in the first place - From the very start, good or bad, there's always been that element of sensationalism that helps to drive interests back to entities that might happen to be feeling less relevant.

Our internet properties didn't suffer when Firefox did this, and I find it extremely doubtful that our internet properties will suffer any great consequence with regard to Google Chrome's new little function either --