Well that’s good news 
2 Likes
I am curious if @irvinewade’s domains have been positively affected since they have been mentioned. Here is the review for my domain name:
The trust score is 21.1 because I have no time to deeply research 359 static site generators on JAMStack, let alone decide on what exact content I want to publish on the domain itself.
4 Likes
Ah, well, I didn’t keep records from 3 years ago - so I can’t directly compare with authority but …
OK, one of my domains is still (unsurprisingly) in the situation of having no Creation Date.
It is ranked 2.5/100, “one of the lowest ratings”, and described as Dubious. Very New. Suspicious
… so that bug hasn’t been fixed. I suspect that the domain’s rating has decreased. This rating is completely useless. It seems it massively overweights what it thinks is a new domain.
All the other domains are rated 52.5 and described as Questionable. Minimal Doubts. Controversial
This seems to be almost a default rating. It means that the “scam detector” lacks discrimination. It gave the same rating to a domain that has no home page as compared with domains that do have home pages.
I find it difficult to reconcile “questionable” with “minimal doubts”, and the assigned rating. Minimal doubts should mean a fairly high rating.
There is absolutely nothing controversial about any of the web sites - and of course all of them are legitimate (but you’ll have to take my word for that 
).
I note that most if not all domains have censored (redacted) registration contact information. This is for my privacy but it may be impacting on the rating.
As with 3 years ago, there is a lack of metadata elements in some of the web sites but others do have metadata. It seems to make no difference. That said, I don’t know what specific metadata elements it wants to see.
Just for a laugh
gnome.org sector: beauty WT? rating: 52.3
filezilla-project.org industry: jewelry WT? rating: 52.5 (and same descriptors as my web sites)
my.gov.au rating: 11.5 Untrustworthy. Risky. Danger.
LOL. I don’t trust my government either but that is because of loopy legislation, garbage governance, and human rights infringements, not because of any dodgy web site / domain. I do actually choose to trust that web site, and use it fairly often over a period of some years, so the rating is basically useless.
I didn’t experience that problem this time. So perhaps that is something that they have fixed in the intervening 3 years.
I stand by what I wrote previously:
I’m thinking that it is of limited use. I would completely ignore the percentage rating and instead look only at the individual items reported.
I would not use this “scam detector”.
It is a pity that they don’t elaborate on more of the “53 powerful factors” that their algorithm takes into account.
As an experiment in automated legitimacy rating … they still have their work ahead of them, in my opinion.
Hey, at least none of my domains was rated as “just a façade”. 
4 Likes
Probably because offering privacy front-ends gratis to the Purism community is neither intended as a business nor associated with a popular industry I am aware of. Here is the review for PINE64:
Their trust score is 65.1, which is 12.5 less then Purism, but their Trustpilot score is very poor at 1.7 stars:
Purism is marginally better on Trustpilot at 1.8 stars:
2 Likes
Just add some Google and Facebook trackers to your sites and I’ll bet the ratings will skyrocket. 
2 Likes
I would rather have my domain name’s trust score become -100 permanently.
2 Likes
That raises the legitimate question of exactly what those 53 factors are. Does third party content make a web site better or worse? Does the rating of the third party itself rub off on the web site’s own rating?
As a final example of how dubious the results are, I checked my employer’s web site (and, yes, again, you are going to have to take my word for it that it’s a legitimate web site and a legitimate company) … score: 8/100 which is absurd.
Let’s check back in 3 years to see whether they have got their algorithm sorted out by then. 
3 Likes
Maybe Australian and other foreign websites are considered untrustworthy due to their geographical region/TLD. Here is the review for Spark, a major telecommunications provider in New Zealand:
Their trust score is 27.4.
1 Like
I did wonder about that. There seemed to be a correlation with the .au TLD and lower scores (not tested statistically) but more careful comparison would be needed to isolate that factor and … I didn’t bother.
geoip is fraught.
My employer’s web site is definitely in Oz but is using a certain evil cloud provider, which can cause confusion about location. My government’s web site looks to be hosted by AWS and based on the ping time I would say is in Oz but is misclassified by some as being in the US. If any web site uses a CDN of some type then the view from the “scam detector” may be quite different from the view here (DNS / anycast routing) - and, during an outage, the location could change unpredictably.
My own web sites are all outside of Oz and one is in .au and the others are not.
As you say, they get the industry sector badly wrong on several domains that we have tested - and that may negatively influence scores as well.
1 Like
“Is Scam Detector legit?”
3 Likes
Here is the review for PureOS:
The trust score is 18.5 due to the Scam Detector claiming the domain to be detected in blacklist engines.
Here is the review for Librem One:
The trust score is 14.7 due to the Scam Detector claiming the domain to be detected in blacklist engines.
Here is the review for Purist:
The trust score is 72.6.
liberty.one is no longer a domain name registered by Purism.
1 Like
All this rating crap! Sounds just as reliable as AirBnB’s, Temu or Amazon’s ratings…
I do my own reputation check this way: upon entering a new website I never accessed before, I look at what NoScript and UBO tell me - and if I see them all cockroaches, Google, Meta and all other malevolent pest thriving, I don’t even try to unblock scripts - I just run away and too bad for them!
3 Likes
I’ll agree. Algorithms are good for indicative purposes but often not good for describing anything complex. In this case, it’s not even indicating properly.
But what I often find is that the base scoring system type is one of the key mistakes to begin with: a “top score” is mostly misleading and measurements like that askew reality when they become a competition (tail wagging the dog) and a necessity (human interpretation of indicativeness changes to the number being “fact”). Any kind of “five stars” or “score of hundred(%)” is like that - eBay, Amazon, ABnB, Uber etc. Sometimes it’s producs, sometimes services, sometimes people/employees. But it’s rediculous to expect perfection (like it exists and isn’t relative) all the time. And such scores do not leave any room for exceptions. Bell curve is more healthy than power law curve. Having “3/5” or “50%” as the expected neutral target that is expected would be better - any deviation from that would be more indicative of… something. It’s usually not feasible to maintain a perfect score - costs too much (not just money but sanity, stress and social aspects too). For this kind of “trust score” a 50% should be the base and if there are strong indications of doubt, lessen score, and if there are supporting factors (say, outside audit or participation in some established peer network), it would be better - but never perfect (as anything can happen and change - nothing stays the same all the time). A “72.6” would be insanely good score in such a measurement system.
So, my thought is, be wary of systems that make the assessed compete like that because it’s detrimental in the long run - and especially, if most can attain (near) perfect score, as that usually means the requirements are weak. Here, of course, there are other more glaring faults.
1 Like