Should we remove numeric scores for reviews?

[StrayKAT]

I should point out when I said "people" need to use review numbers correctly.. I just meant general trends. Not necessarily here. I just realized it sounded like I was attacking this place specifically.

 

[GlacierNine]

I don't even slightly care if you want to give scores or not - What I care about is when those numbers mean exactly jack shit.

Go through your own review archives. Tell me what the "average" number is for any review. Pretty fucking high, isn't it?

Rather than worry about arbitrary numbers assigned by reviewers, I'd much rather see you get Darksaber to test thermals and noise. Now *THAT* would improve things.

 

[cadaveca]

Finishing with an overall review score, at least that way strengths and weakness are clearly identified at a glance on the conclusion page, this leaves someone who priorities noise in a small form factor build to concentrate on temps and cooling (as maybe this is where it may score particularly well) and be less concerned with say price/value, the same would apply to an enthusiast who is prioritising performance and cares much less about power consumption, that way a good review overall may be an excellent review for the enthusiast if you get my meaning.

Go to the main page. Click reviews. See the scores? How do we relate that there in an easy-to-understand fashion?

I don't even slightly care if you want to give scores or not - What I care about is when those numbers mean exactly jack shit.

Go through your own review archives. Tell me what the "average" number is for any review. Pretty fucking high, isn't it?

Rather than worry about arbitrary numbers assigned by reviewers, I'd much rather see you get Darksaber to test thermals and noise. Now *THAT* would improve things.

There is much I could say about this particular subject... and maybe that's part of why W1zz posted this poll in the first place. I've even written about it in reviews, too.


I also commonly say, if you don't see it on these pages, don't buy it. It might be nice if you bought it from me directly, and I got a cut of all those sales, too. But that might make for bias?

 

[Tatty_Two]

I was not suggesting I had all the answers! I just like the weighting element I think for a (semi) enthusiast site it just prioritises attributes to form the overall score in a somewhat bias but pertinent way.

 

[Joss]

I agree with some others in that scores should punish when needed.
Take the latest RTX reviews, everything is above 9. How can you justify those scores when prices are, well, obscene. Personally I think the 2080 FE shouldn't deserve more than 6.

 

[GoldenX]

Nah, the card is great, but a scoring system in categories would make the overall score a lot lower than 9 thanks to pricing.

 

[cadaveca]

I was not suggesting I had all the answers! I just like the weighting element I think for a (semi) enthusiast site it just prioritises attributes to form the overall score in a somewhat bias but pertinent way.

Aw man, I was hoping you would! LOL.

NO but really, perhaps if we can find a way to do things clearly, then we can convince W1zz to adopt it. Lots of little details to cover.


Yet with only 67 votes out of 80000+ members... sigh. Guess i'll bump this up the list...

 

[Flanker]

IMO a singular score isn't that helpful. I would give the Turing cards 9+ in all aspects except 2/10 in terms of value

 

[eidairaman1]

I also saw the grading system change from when I was in school, much as @MrGenius put forth. It changed to the more draconian method when I got to college.

My kids grew up only knowing that 70 or lower was an F.

70 was a D or C, 69 was an F

 

[StrayKAT]

IMO a singular score isn't that helpful. I would give the Turing cards 9+ in all aspects except 2/10 in terms of value

That doesn't sound bad at all.. considering the controversy.

I'd give Optane drives a 9 too.. but take away a point for value/power consumption. Maybe even an 8? This is probably where a Samsung could beat it out on overall points....even though Optane is the fastest and/or most promising SSD.

 

[PLAfiller]

I don't really mind the score. I just read the Pros and Cons on products I am not going to buy. For those I consider buying, I am like a mole digging through unboxings, user reviews and professional reviews. With or without the score in the end, I am not really bothered. In the end I somehow expect to see some number though, may be its a habit or I am disciplined this way from numerous other reviews. Everybody uses scores I guess.

 

[notb]

40+% saying that they want scores removed, well, if you do statistics and assume the people that responded represent the entire readership, that's like WELL over a million readers.

They don't.

Only members of the forum are able to vote, which is a very specific group - more interested in gear and, most importantly, willing to spend more time learning about gear. It's plausible that a simple rating system is of less interest to them.
Moreover, people that voted are active members of the forum, so it's likely they don't even represent the community very well.

You guys have to decide if reviews are for the general public, which also includes people that aren't very interested in PC parts. They might even not understand a large portion of a review.
People want to buy a GPU. They open a site they trust, they read a few reviews and they compare.

I'm sure you're sometimes reading reviews (or user ratings at online stores) of stuff that you have to buy, but aren't that interested in. Cars, bikes, kitchen stoves, backpacks, women's perfume etc. Do you find ratings useful then?

 

[dorsetknob]

people that voted are active members of the forum, so it's likely they don't even represent the community very well.

Probably there are only around 400 to 500 members that are Active enough to post Regularly and they keep the place alive and kicking

 

[cadaveca]

I'm sure you're sometimes reading reviews (or user ratings at online stores) of stuff that you have to buy, but aren't that interested in. Cars, bikes, kitchen stoves, backpacks, women's perfume etc. Do you find ratings useful then?

Absolutely not useful for me. I make these scores myself, so I understand the psychology behind them. My actual opinion here isn't really relevant. It might be a case of "knowing too much"

 

[qubit]

I was not suggesting I had all the answers! I just like the weighting element I think for a (semi) enthusiast site it just prioritises attributes to form the overall score in a somewhat bias but pertinent way.

You mean you don't know everything?! Son, I am disappoint.

 

[RCoon]

Use the same scoring method as JohnnyGuru. Separate your 10 points into five categories (e.g. Power Usage, Gaming Performance, Price, Cooling, Overclockability), 2 points each, and deduct and award based on those five areas. Your final score is thus a little more useful, you can show the scores for each respective area and then provide an overall score for those that only want a big number.

 

[neatfeatguy]

A review is nothing more than a personal opinion of the author's experience. The author writes the review, puts down their thoughts and then grades their experience. Folks need to do more than just stare at a number and jump to conclusions of the review. They need to read through, consider the information and also realize the review is just the author's opinion based on their experience. After they do that, they can better understand where the number score came from and then they can make their own decision of item based on what they know about themselves and whether they can agree or not with what was in the review.

I vote, keep the number score. If the people reading it can't figure out why the score is what it is, then there is nothing you can do to fix ignorance. You can't fix stupid and you shouldn't cater to those that simply cry because it doesn't fit their needs.

 

[GlacierNine]

A review is nothing more than a personal opinion of the author's experience. The author writes the review, puts down their thoughts and then grades their experience. Folks need to do more than just stare at a number and jump to conclusions of the review. They need to read through, consider the information and also realize the review is just the author's opinion based on their experience. After they do that, they can better understand where the number score came from and then they can make their own decision of item based on what they know about themselves and whether they can agree or not with what was in the review.

I vote, keep the number score. If the people reading it can't figure out why the score is what it is, then there is nothing you can do to fix ignorance. You can't fix stupid and you shouldn't cater to those that simply cry because it doesn't fit their needs.

That said, what's the point of a reviewer "grading their experience" if they use a scale that either makes no sense or a scale that doesn't seem to have any real meaning?

There has *never* been a Motherboard review on TPU that received a numerical score of less than 6.5.
No graphics card has ever recieved below a 7.
No case has ever received below a 6.2.
No PSU has ever received below a 7.3
No headphones have received below a 7.9
No CPU cooler has ever received below a 6.5
No RAM has ever received below a 7
No keyboard has ever received below 6.8
No "Other" product has received below 7.2
No Mouse has received below a 6.
No SSD has received below a 7.5
No NAS has received below a 7.5
No external HDD has received below a 7.2
No CPU has received below a 7
No USB stick has received below 7.4
No mouse pad has received below a 7.3


I could keep going through all of the categories of review TPU has done, but I think you get the point by now.

In fact, the lowest score TPU has *ever* assigned to a product was a 6, an honour given to this waterblock in 2017 -
https://www.techpowerup.com/reviews/Bitspower/Summit_EF-X/
and to this mouse in 2007 -
https://www.techpowerup.com/reviews/OCZ/Equalizer/

The conclusion of that waterblock review reads "There are some silver linings in this dark cloud" and "Overall, even without the QC issues considered, there are simply too many options available that cost less and perform better. As such, I can not recommend the Bitspower Summit EF-X as it is now."

The first sentence of the mouse review conclusion is "The OCZ Equalizer disappoints in almost every possible way, right from the tracking performance to the driver setup. "

I'm not against TPU grading products with an easy-to-use number, but what the fuck is the point of a site that literally seems unable to review a product badly, even when the conclusion of the same review clearly outs that product as being utter garbage?

 

[R0H1T]

Well I guess that's the point of this poll, pros & cons IMO are more objective in highlighting a given products' strengths or weakness ~ points/scores on the other hand are a lot more subjective & hence ought to be scrapped.

 

[notb]

Absolutely not useful for me. I make these scores myself, so I understand the psychology behind them. My actual opinion here isn't really relevant. It might be a case of "knowing too much"

Ratings of washing machines are not useful for you? When you need a one, you're quickly studying the field and becoming a washing machine expert?

 

[GlacierNine]

Well I guess that's the point of this poll, pros & cons IMO are more objective in highlighting a given products' strengths or weakness ~ points/scores on the other hand are a lot more subjective & hence ought to be scrapped.

I don't really think "subjectivity" is responsible for the fact no review in TPU history has ever been given a negative numerical score.

If I were a conspiratorial type, I might suggest that actually, the reason for that is the fear of no longer being sampled for new products if a company were to take offense at their garbage product being given a 3.

 

[notb]

There has *never* been a Motherboard review on TPU that received a numerical score of less than 6.5.
[cut]

You've based this opinion on an assumption that the rating scale has some possible values below 6.5 (like 0-10). Maybe it doesn't?

Or maybe it's just very unlikely for a product to get <5?
You know... if "5" means that something just works, but is unattractive and very mediocre, then scores below 5 would be for products that don't work.

Like, you know, Chinese knockoff keyboards that disintegrate after 2 months.
But TPU is not reviewing such products - they mostly review high-end, expensive (almost luxurious) gear.
Let's look at the keyboards, since I've already mentioned them. There is no review to date of something really cheap and basic - like A4Tech.
There's only a single Logitech keyboard - CRAFT (the most expensive one they have today) - despite Logitech being the most popular brand and having many gaming models in the lineup.

tl;dr
TPU doesn't really cover bad products, so why would they have to give bad marks?

 

[R0H1T]

I don't really think "subjectivity" is responsible for the fact no review in TPU history has ever been given a negative numerical score.

If I were a conspiratorial type, I might suggest that actually, the reason for that is the fear of no longer being sampled for new products if a company were to take offense at their garbage product being given a 3.

Well there's that but then you'll rarely, if ever, find an objective review using that scale on the internet.

 

[EarthDog]

As I have said numerous times before over the years....absolutely remove it because you cannot stand behind scores empirically. What is the difference between an 8 and 9? How about 8.5 and 9? Whole numbers are frowned upon....the fact that there are decimals in use makes it even worse. Scores are ok if you have a set system in place to 'dock' points. But since this is purely subjective, the numbering scheme is method is pretty useless (to me) and means very little. Dave's post is quite telling that the scores are seat of the pants and feel (butt-dyno).

The thing with JG reviews is that he takes off points for the same things so there is an empirical base there. It has nothing to do with the categories.

 

[Joss]

Use the same scoring method as JohnnyGuru

Yes, or something similar.
Something that would produced a more balanced final score because, I'll say it again, TPU's latest NVIDIA's scores are way too high.