New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
Working on this now.
Hi @Alyx ,
It is of my concern as well.
I suspect the ServerVerify team may have accidentally generated around 4500 pages of AI reviews for real hosting companies.
Mistakes like this can happen to anyone so I'm just bringing some attention to the issue.
Most usernames follow the pattern of
{name}_{surname}{3-6 digits}or just{nick}{3-6 digits}.Looking up the review contents or the names on the web doesn't yield any results.
Usernames that don't follow the pattern are probably the only real reviews on this site.
Reviews in some cases are anonymized for privacy.
However you raise a good point here that needs attention. We are working on this now.
Sometimes I accidently generate 4500 pages of reviews as I grunka.
Edit: Mate rick confirm it pin it reakt it.
Anonymized as in AI generated?
Might be nice to also show the model of the server
What would be the model? What do you mean?
95%+ of them are obviously AI generated, and not even that good of an AI, you might need to have a look at your data source.
It's not just one AI, it's at least 2 different versions. There's a very big writing skill difference between the AI that wrote the "randomword[number]" and the AI that wrote the "randomword_randomword[number]" reviews.
Whatever the provider calls that particular configuration
Yeah I agree, we are working on it right now. Going to get it fixed right up.
The idea was to anonymize the reviews and avoid content duplication. We are handing this with top priority now.
They shall be listed on a per-plan basis.
I am struggling with the concept of an anonymized review. Either someone has reviewed a host and you have permission to publish that, or it's not a review and it shouldn't be presented as a review. I am assuming you are taking legal advice here (I guess providers could get upset by seeing negative AI-generated fake reviews).
should have just said they were placeholders @Jbiloh but jokes aside, good luck with it
No, I can't and I wasn't involved in that project in any way, shape, or form.
Nor would I even think about using an AI to create fake names for user "reviews" in order to "anonymize" them, because a review without an actual, and preferably known in our community, user behind it IMO is less than worthless.
Yes and no. On the one hand single core performance of course is important, often even decisive, on the other hand though it is useful to know what kind of performance can be expected from all cores. That's why in my benchmarks you get to see both numbers, plus a few more (e.g. crypto) for processor and memory.
it shows you werent involved
Great👏👏👏
Thanks
We put a ton of work into this.
We'll keep making it better over the next few days, weeks and months.
In addition to all this great feedback we've gotten so far what would be helpful is if folks start processing benchmarks plus leaving reviews.
Congratulations. The idea is good.
Good luck and allow me to wish you success.
I don't know about that one. You took something great about a community freely participating in giving reviews and benchmarks, then turn it into a source of income. Now you want us to leave more reviews so you can make more money out of it, like loading it all with ads.
Yeah... good luck with that one. Personally I freely provided benchmarks and opinions for other possible customers, not for your profits.
As a side note, I want to see a review about ColoCrossing's data breach which was the biggest drama of this year so far, but something tells me we won't see that. So yeah: good luck.
The request was more about testing and exercising the system.
The idea behind ServerVerify is to help aggregate a bunch of information in a single place to help buyers make better purchasing decisions plus help motivate hosting providers to improve the quality of service/performance they offer customers.
Initially yes, but then comes the profits. I know how it works, and I dare to guess others do too.
The idea is great, but the reasoning behind it might not be the purchasing decisions, instead it could simply be your profits.
This is your flaw. In your huge desire for profits you failed to see the passion of providers delivering $7/year deals. You fail to see providers not doing deals for profits, but for the community. A provider with $7/year deal will not have great benchmarks on those deals, precisely because such deals are oversold; and even though that same provider might offer the same plans with bigger price and greater quality and even better support, it won't have great benchmarks on your website, because the aggregated data will exist more for the low-end plans.
The beauty of this community is also the hype which generates SEO, so the exposure will be for the advertising towards selling more expensive plans. A provider's image does not rely merely in benchmarks and selective reviews, but also in the hype, in the listening, and the giving back to community, which can't possibly be reflected in a website like that. From what I can see, you wish to take away the effort of reading forums, and just summarise it all in a website with some stats and reviews, all for some worthless profits.
Let's take ColoCrossing for example: great benchmarks, but it had a huge data breach which they did not even take responsibility for publicly. I wonder how you would reflect that on your website. My guess: you won't. And don't even get me started on the IPv6 issue.
Good luck with your sales!
@default so you are aware ColoCrossing has been deploying IPv6 upon request for the past couple of months. If you have service there you could ask ColoCrossing for IPv6 service.
The algorithm behind the scoring is a bit flawed in my opinion.
Since YABS doesn't have the same parameters across the board, hence weightage should be given more towards single score GB6 than multi.
Currently, let's take an example.
Evolution host with 7950x has multi score of 11000 with 16 cores if I recollect correctly.
Bero on the other hand is at 9000 odd with 4 cores.
In above, evolution has a score of 53 while Bero has score of 44. This is very misleading towards buyers and it will be even better if all the servers are NOT ranked but only give comparison in tabular structure where people can filter out the way they want.
This score system will be very controversial in future and will be a weapon to abuse in order to acquire more customers by manipulating the parameters.
There seems to be a lot of inaccurate information on the site.
For example: https://serververify.com/benchmarks/49ad1b86-1181-4d90-a0cb-b401d45be1df
It shows that it's a benchmark for Verasel, submitted by fiberstate, but this is not true
Or this one:
https://serververify.com/benchmarks/5312f8cc-fce9-47c5-b624-81317d6ada42
It shows that it's by 1GServers but this is very clearly a benchmark for Advin Servers.
https://serververify.com/benchmarks?isp=Advin Services LLC&page=1
Quite a few benchmarks are mislabeled as being from 1GServers, Scaleblade, or MassiveGRID instead of Advin Servers. How did it even come to this conclusion?
Or for example:
https://serververify.com/benchmarks/39b8e9f4-0a70-4400-846d-779dbc9b61fd
This one is labeled as being from Advin Servers though it appears to be done by Layer7.
It's simple really, @jbiloh bought the $30 Claude plan and banged this out in a month in between the rate limits and the 72,000 auto generated AI reviews 'anonymized' reviews before the subscription expired.
Anyways, I threw in one of our idling servers for fun:
https://serververify.com/benchmarks/dd55001c-56d8-4204-a15e-c567209123f2
Maybe because I am not involved in the development
Idling 128 cores is a crazy flex.. any ServerVerify launch day deals to celebrate?
I've created a promo code called beanman109 that applies to any product on our site to celebrate the launch of ServerVerify
It's broken, so that's why it's idling
When fat32 promo code