Testing improvements to the rating & review system

[edc_lab]

58 minutes ago, emmaki said:

Oh, you sweet summer child...

Also, it's really obvious that this isn't your perspective, but what ChatGPT thinks. Fiverr's not particularly listening to humans sharing their opinions. What makes you believe Fiverr will take ChatGPT's vague ideas into consideration? Especially when it has its own AI systems with more insight than man or machine to inform their data analysts of the good, the bad, and the ugly of this test? Analysts who likely have different questions to answer than those shown by sellers on this thread. 

I noticed a lot of the recent amazing templates/emails etc. for sellers from Fiverr were equally poorly created with AI, so you're in company. I'm always surprised at how poorly AI is being "adopted" by most people when there's much better use cases for it. Have you seen Google search results lately? Most of it is badly-done AI keyword spam. Even the news media is in on it with BERT. And of course, there's the AI infestation on forums where users think they might gain in some way - usually financially - by using AI answers (Quora is particularly bad for this).

The point being, if you're going to introduce something as your perspective, then it should be yours, especially if you're going to tie your perspective into your service offerings on Fiverr. 

My opinion has been simple for years: if Fiverr sellers want to see change and actually get listened to - including those who actually get flown into TLV for chats with the CEO and now, 6 months later, are packing up their bags to leave due to the stress of Fiverr's increasingly draconian metrics - they should all simply go out of office for a few months. Fiverr is pretty hopeless without sellers (or buyers). That'll pan out in two ways: 1) Fiverr actually listens, or 2) AI workers are deployed (cheaper to run, less tiresome, no commissions etc). 

Won't happen though. You know why? Because no matter who leaves, there are always new sellers coming in, with varying levels of competency. Hence the increasingly draconian metrics.

One might ask a simple question: why would Fiverr need its metrics to get ever-tougher, more opaque, and more reliant on AI at the same time it goes upmarket? 

The second question that follows is naturally: where is the balance in a system that treats all sellers exactly the same and still uses systems that were put in place when the site was a lot smaller?

The third question: why has Fiverr never really responded well to sellers saying "wow, that's bad" to them about these systems? Back in 2017, I told the CEO in TLV that the Levels system was a horrible idea. He disagreed, of course.

I know they listen - but I also know they rarely implement. Anyone can listen to people talking. It ultimately means nothing. It might make the person being listened to feel special and validated, at least until later, when they realize that they were not heard. 

This whole post is a monument to Fiverr's ability to listen, with a few comments here and there being "passed on" to the team. 

They do listen quickly when you tell them that one of their glossary items is outdated though. They removed that glossary item. But not the other obvious outdated ones... imagine if a Fiverr seller were to do such slap-dash work for their client 🙂 Take a look at the Early Payout page in the Help Center. The description of how it now works is badly-written to the point of nonsensical and contradicts the Forum post on the update, which promised the team would "update" (presumably here) on Monday. It's now Tuesday evening. That's corporate listening and action for you. Of course, the only people interested in that topic are the people who like Early Payout, who are full of profuse thanks to give the corporation even more of their money to get their money. Which I find amusing. 

tl:dr: it's must more interesting when you share your actual perspective, just like I did, all with the laborious use of my fingers and brian. ChatGPT is not interesting. It is an increasingly-nerfed tool, not your mouthpiece. Please don't confuse the two. One is powerful. The other is not. YMMV on whether any of this wall of text was interesting though, and that's fine. 

First of all, thank you for taking the time to respond to my message with such a lengthy reply (I know, this sounds very ChatGPT, but it's what I think).

Your questions are all valid and that's precisely what I meant in my previous message, listening and understanding the perspective of those with "on field" experience. Collecting this feedback can help optimize and highlight not just their profits but also the individuals generating these earnings and adding value to the platform itself thanks to their professional services. Of course, it's fair to look at metrics and data to try and optimize profits, but it's also crucial to listen to those creating these gains and try to implement their suggestions.

Allow me to conclude with a note.
From your message, I felt a clear resentment toward the improper use of ChatGPT, and it saddens me to think that in just a few lines, you've formed a completely incorrect idea of me. Essentially, you've placed me among those who abuse these tools to sell email templates (you basically associating me with pseudo-scammers), and I really dislike this. I want to clarify that what I write are my own thoughts and if I use ChatGPT it's solely to correct any grammatical errors made while writing directly in English.

You mentioned that I don't use my fingers and brain like you do, but the truth is I'm making a double effort, as this isn't my native language. So, I apologize, but you've greatly misunderstood that aspect.
My aim has always been to contribute to constructive discussions, so the note I'm making is respectful, and I really want to point out that.

[donnovan86]

30 minutes ago, visualstudios said:

You're missing the obvious point. Someone who already has 10k 5 star reviews, under the new system, will have an insurmountable advantage over someone starting on this new system with 10, 20, 100 reviews.

I am in that 10k+ review club with one of my gigs. Since all those reviews were accumulated due to lots of blood, sweat and tears, they are valid reviews. You can't just dismiss them because others chose to either go for a lower price, less orders approach. I respect that, but you can't ignore either side of the coin. That's why I think it's tough for Fiverr to try and equate those older reviews to the new system.

But hey, they went from up/down to 5 stars, that was a huge leap. 

Although people are barking at the wrong tree here. Fiverr buyers can actually see what the review score is before they press Publish. So there are 3 things here:

• They don't care and randomly press emoticons
• They rush to press stars, see the review score and press publish.
• They knowingly choose to write that review and score you like that. 

I get it there can be mistakes but come on... if they show the review score BEFORE PRESSING PUBLISH, how can people say buyers are misguided to press stars and 4 star reviews are a mistake, when they clearly chose that. Even if they made a mistake, they can revert and not press publish so... I think the problem stems from buyers rushing or not caring. If they wouldn't show the review score before publishing the review I would understand, but a buyer can see that so.. I still believe a lot of buyers randomly rate people, speaking from my own experience here.

[catwriter]

3 minutes ago, edc_lab said:

if I use ChatGPT it's solely to correct any grammatical errors made while writing directly in English.

According to an AI detector, your previous post was 99.99% AI generated.

This post of yours isn't AI generated (according to the same detector).

If you're going to use ChatGPT to express your thoughts, as you did in the post I've mentioned, please at least keep the forum rules in mind. And the rules state that, if your post is AI generated, you should mention it (cite the source).

[priyank_mod]

41 minutes ago, visualstudios said:

You're missing the obvious point. Someone who already has 10k 5 star reviews, under the new system, will have an insurmountable advantage over someone starting on this new system with 10, 20, 100 reviews. The 10k review seller will be able to maintain the 5 star average for years, therefore getting an unfair advantage over everybody else. Doesn't matter if they're not better, and if the reviews they're getting now aren't better. Their profile and gigs will show 5*, where everybody else's will not.

No, no. I have not missed out on this point at all. Multiple people have expressed it on the thread and it is a valid concern. 😇 I had asked Kesha on day 1 itself - how they will merge old ratings with the new?? Till they provide a concrete and fair migration roadmap to the future ratings system - we can only assume/hope about things. 

I received my 1st rating under the new system today - it is a 5-star on all 3 new parameters but on the profile it shows 5-star like the current/soon-to-be old system. 

Keeping fingers crossed about the way forward!! 🥲

[visualstudios]

18 minutes ago, donnovan86 said:

I am in that 10k+ review club with one of my gigs. Since all those reviews were accumulated due to lots of blood, sweat and tears, they are valid reviews. You can't just dismiss them because others chose to either go for a lower price, less orders approach. I respect that, but you can't ignore either side of the coin.

Exactly. And that's why you can't have the cake and eat it too. If it was easier to get 5 star reviews under the old system than the new one, the system is unfair for those who had a ton of previews reviews. This is obvious.

If, on the other hand, the difficulty of getting a 5 star review on both is equal, then obviously there's no issue. But then why change it? They change it purposefully, to drop review averages. I may be wrong, time will tell, but I'm willing to bet on it.

[edc_lab]

18 minutes ago, catwriter said:

According to an AI detector, your previous post was 99.99% AI generated.

This post of yours isn't AI generated (according to the same detector).

If you're going to use ChatGPT to express your thoughts, as you did in the post I've mentioned, please at least keep the forum rules in mind. And the rules state that, if your post is AI generated, you should mention it (cite the source).

Hello @catwriter thanks for precising that! 
I want to point out that I didn't use ChatGPT to write the core concept of my thoughts. However, this time, given the delicate nature of the topic, I asked to fix grammar errors to avoid silly mistakes.
Please keep in mind that gathering data, listening to feedback, and creating solutions around the end user are the basics of UX Design, I didn't write anything exceptional.
But again, thanks for the hint, perhaps you all appreciate more genuine mistakes over perfect texts (at least here on the forum, not in some articles you can find online ).

[visualstudios]

22 minutes ago, donnovan86 said:

I get it there can be mistakes but come on... if they show the review score BEFORE PRESSING PUBLISH, how can people say buyers are misguided to press stars and 4 star reviews are a mistake, when they clearly chose that. Even if they made a mistake, they can revert and not press publish so...

Then the people on this thread who claim that they got 4 star reviews when their buyers gave maximum ratings everywhere, and even left tips, are lying. Or their buyers are, although that wouldn't make a lot of sense given they left tips. Interesting.

Edited December 19, 2023 by visualstudios

[donnovan86]

Just now, visualstudios said:

Then the people on this thread who claim that they got 5 star reviews when their buyers gave maximum ratings everywhere, and even left tips, are lying. Interesting.

I think the problem here is a lot of buyers just randomly press buttons. I know from my experience they do, and this new system most likely caught people offguard, they didn't understand or care, didn't study anything, didn't see the 4 stars at the end, as you saw Fiverr automatically shows the review score before you post. So buyers definitely know what they are rating you. Whether they care or not, that's another thing. Obviously if you ask any buyer, they say it's Fiverr's fault. But again, as I said, how can you state that when Fiverr clearly shows you the review score before you press Post Review.... It's not like they change it once you post it. You can clearly see what review score you are posting as a buyer and by writing the review and pressing Post, you agree to it.

 

[visualstudios]

11 minutes ago, donnovan86 said:

I think the problem here is a lot of buyers just randomly press buttons.

Well, they didn't before. They left 5 star reviews before, they didn't seem to be pressing buttons randomly then. If this new system makes them press buttons more randomly, it's clearly a worse system.

Just looking at your profile shows it clearly - you have over 18.9k reviews, so it's statistically very relevant. Of those, over 18.2k are 5 stars. If a relevant percentage of buyers were rating things "randomly", your profile would be the statistical anomaly of the century. It's not "a lot of buyers randomly pressing buttons". That may happen, but it's very rare. At least in your case, it must be, otherwise the rating distribution would look very different.

Edited December 19, 2023 by visualstudios

[emmaki]

34 minutes ago, visualstudios said:

You're missing the obvious point. Someone who already has 10k 5 star reviews, under the new system, will have an insurmountable advantage over someone starting on this new system with 10, 20, 100 reviews. The 10k review seller will be able to maintain the 5 star average for years, therefore getting an unfair advantage over everybody else. Doesn't matter if they're not better, and if the reviews they're getting now aren't better. Their profile and gigs will show 5*, where everybody else's will not.

If they got rid of the old scores, and everybody started from 0, then the new score average would make sense - everybody would be lower. But that's not the case.

Like, if the system had always been like that, since the beginning, nobody would be complaining. There are plenty of platforms where a 4.0 star average is good. What you can't do is condition people for years that it's 5 star or "bad", and then switch it. That doesn't work.

To be fair, putting everyone back to 0 ratings would just create a huge storm of brown stuff. 

Anyway, I remember when Fiverr used to show when people had over 10,000 reviews. They reduced that to 1K+ whenever. And then they did the metrics thing so that if you don't do anything on Fiverr for ~60 days, your profile shows as 0 stars (this happened to me earlier this year lol) so that's already inbuilt and why you're not meant to vacay (I was not on vacation: I was sick. With lots of savings for this kind of thing, because you know, gig economy workers have no protections...) for more than 2 months... 

Fiverr has just created a system where you need to bring your a-game every day for as many orders as possible. The end. Doesn't quite match up with the whole "going upmarket" thing where order volumes are naturally lower, and you can see the results in the multiple comments from long-standing Pros/TRS/whatever either planning to leave or roll back their efforts on the platform in 2024.

I'm sure they're all quite aware they're wholly replaceable, but isn't the real question "for how long"? Fiverr doesn't have a great image with professional communities in many industries and the restrictive contact policies probably discourage others. The current system rewards volume (or agency) work... and it's hard to see where emoji fun reviews fit into any of this. 

Like, what happened to Paul Maplesden? He was such an enthusiastic Pro, wrote blogs, evangelized about Fiverr, active forum member, then he just quit and never came back. Who frustrated that champion away from the platform? Sure, it could be the clients... but what if it was simply Fiverr, the Great Listener? 

Or take VoiceoverPete. I won't go into detail here, but he was essentially told what he was doing was OK then thrown away in the blink of an eye...

And over on Reddit, there are more than enough people recommending to only use Fiverr to "get started" then move on ASAP. 

Doesn't this eventually leave Fiverr with a "Pro" marketplace populated by people at the start of the career at some point? The downgrade already happened once. It's just a question of "when" now. Perhaps Certified fixes that as the new "Pro" marketplace, but that's essentially just a glorified outsourcing 'expert' helpdesk for other huge companies with bad CS. Fiverr certainly doesn't disclose publically that the certification is only as good as the "partners" make it (and I shall leave that at that - IYKYK).

None of this really squares with Fiverr wanting to go upmarket. If anything, it only drives home the point that Fiverr is essentially a content mill. With emojis. That listens. Sometimes. While apparently claiming to be transparent about a review system that downgrades reviews and makes no effort to come to the community and say "oops, little technical issue lol" when it is raised, several times. 🙂

Unless the cash stops coming in, ain't nothing going to change and we're all going to like what we're given or move on (by choice or AI disapproval). 

Just now, edc_lab said:

First of all, thank you for taking the time to respond to my message with such a lengthy reply (I know, this sounds very ChatGPT, but it's what I think).

Your questions are all valid and that's precisely what I meant in my previous message, listening and understanding the perspective of those with "on field" experience. Collecting this feedback can help optimize and highlight not just their profits but also the individuals generating these earnings and adding value to the platform itself thanks to their professional services. Of course, it's fair to look at metrics and data to try and optimize profits, but it's also crucial to listen to those creating these gains and try to implement their suggestions.

Allow me to conclude with a note.
From your message, I felt a clear resentment toward the improper use of ChatGPT, and it saddens me to think that in just a few lines, you've formed a completely incorrect idea of me. Essentially, you've placed me among those who abuse these tools to sell email templates (you basically associating me with pseudo-scammers), and I really dislike this. I want to clarify that what I write are my own thoughts and if I use ChatGPT it's solely to correct any grammatical errors made while writing directly in English.

You mentioned that I don't use my fingers and brain like you do, but the truth is I'm making a double effort, as this isn't my native language. So, I apologize, but you've greatly misunderstood that aspect.
My aim has always been to contribute to constructive discussions, so the note I'm making is respectful, and I really want to point out that.

I don't know you in the slightest, Ed. I just know that you used AI in that comment. I'm simply saying that if you're going to share "your perspective," make it yours, grammatical mistakes and all: 

If you want to go down the route of "you don't know me" then I can only agree, but point out that I know two things about you now: you use AI - no big deal - and that you deny using it except for a very mild use case. Nobody here can know how extensively or not you used ChatGPT in your comment, but let's look at a use-case where it may harm you:

You write a report for a customer using AI in full or in part. The customer notices it is in GPTese. They are one of the AI luddites. You get a terrible review, emojis or not, because as far as the customer is concerned, you simply got a machine to do the work: something that they could have done for free. This isn't an outlandish scenario. It's what is happening today on Fiverr. 

Now consider how using AI might impact people's decision to order from you. If AI can do it for "free" - or more importantly, they think they can get it done just as well for free - why should they pay you $5, $10, $20? 

I use ChatGPT and Perplexity all the time and I do not resent AI - I'm just tired of people abusing it and, all too often, denying it when caught red-handed. It's a great workflow tool, but a terrible writer and researcher. The templates, by the way, I mean Fiverr's own staff are sharing them on the Forum...doubtless hundreds of sellers are using them right now, some may even think they're "better" than ChatGPT because a trusted person provided them, and, of course, the English is "better".

Are you so quick to jump to the "I'm not a (pseudo)-scammer like them" defence now?    

It doesn't matter if your English isn't perfect. Believe me, I live abroad and I end up using translation tools all the time in situations where I have no idea what to say and I murder the (ridiculous) grammar on a daily basis. I get it. I know I sound like a complete idiot and don't the fact I cannot articulate myself as I would like, but that is just the nature of learning languages until fluency and beyond. Either way, a better use of ChatGPT to help you articulate your thoughts would be to get GPT4 and ask it to speak to you in English and practice conversations. It will correct you as you go along. 

My point remains the same: If you're going to say "personally my opinion is I think" or something along those lines, use your own words - not the top-notch writing you extract from delving into the realm of ChatGPT. AI  has never been the problem. Like any other tool, how people use it is the problem.  

(Just use Grammarly basic for correcting your spelling and grammar). 
 

[donnovan86]

6 minutes ago, visualstudios said:

Well, they didn't before

Not in your case. I tend to deal with a lot of volume work, and I can tell you for a fact buyers tend to rate randomly. It might not always be the case, but they definitely do in a lot of situations. 

7 minutes ago, visualstudios said:

If this new system makes them press buttons more randomly, it's clearly a worse system.

I mean, they just have to press 2 emoticons. The stuff under those emoticons doesn't influence their rating. But I assume some buyers are confused and they think that also affects the rating, which Kesha said it clearly doesn't. And again, you can see the review score at the end before you publish it. So willingly or unwillingly, buyers agree with the review they are sharing, and they see the score. That's why I think people randomly place stars, and the emoticons maybe started confusing them even more. If a person is careful when rating, they would see the review score they are publishing, right? It goes back to what I said, from my personal experience I know a lot of people rate randomly. They just get the work and move on. Some buyers take the review process very seriously, but there are many others that don't.. again I know that for a fact from my personal experience. And I had $5, I had $200 orders, so pricing doesn't factor in as much. 

[visualstudios]

7 minutes ago, donnovan86 said:

I tend to deal with a lot of volume work, and I can tell you for a fact buyers tend to rate randomly. It might not always be the case, but they definitely do in a lot of situations.

Just looking at your profile shows it clearly - you have over 18.9k reviews, so it's statistically very relevant - the larger the sample size, the more relevant it is.

Of those, over 18.2k are 5 stars. If a large percentage of buyers were rating things "randomly", your profile would be the statistical anomaly of the century. It's not "a lot of buyers randomly pressing buttons". That may happen, but it's very rare. At least in your case, it must be, otherwise the rating distribution would look very different.

What you've written there can't be true. It's math, simple as that.

Unless you're saying that a lot of people rate things randomly, and by sheer luck you have 95+% of those people picking 5 stars - where if the distribution were truly random, we'd expect 20%. That's amazing.

Edited December 19, 2023 by visualstudios

[adleypromo]

Hi @Kesha

 

I think you are violating one of major rules in UIs. You should not do major UI changes suddenly. It should be a gradual change. Otherwise returning buyers will get confused with new review system and accidentally they post reviews which they doesn't mean to. I have faced several incidents like this in last few days where 3 of my returning buyers who are working with me for a long time had accidentally rated low stars in the review and when I inquire whether there was any issue with the service all of them said they haven't meant to put low rating, it is this new reviews system which confuses them.

 

Also by adding several steps in review process, you are making it a complex process and also this tends the buyer leave without leaving a review. The best review system would be 1 single page contain star review system for each criteria you are measuring and under that leave a box for private review.

 

So please consider this point

Edited December 19, 2023 by adleypromo

[donnovan86]

Just now, visualstudios said:

What you've written there can't be true. It's math, simple as that.

Well, one of my bad reviews is a person who clearly wrote good job and told me the work was great, yet he left 1 star. It was not a $5 order, it was a $50 order. Doesn't matter, he randomly left stars, which supports what I said above.

Multiple 4 star reviews are from people that did the same mistake you and others are saying regarding this new system, they thought they gave 5 and everything was fine. Again, randomly pressing buttons.

Multiple 3 star reviews saying: it's fine, he already published it, but he was expecting $100 worth of work in a $5-$10 article and stuff that was never included in the gig anyway. In the grand scheme of things it doesn't matter, but it's very frustrating to encounter these issues nonetheless. 

Having the same issues happen with this new system doesn't surprise me, because it was already happening in the past. I think what buyers misunderstand are those options under the emoji. But as I said, how can you say it was a mistake when you clearly see the review score at the end.. That's what I don't understand. It's either negligence or people willingly rate sellers like that. I get it one or two people might make a mistake, but so many cases.. come on. 

I am not saying this system is perfect, I am not a fan of emojis on a professional platform. But disregarding that, a buyer still gets to see the review score before publishing. So I just don't understand these "seemingly" mistakes. Because as a buyer, you see the exact review score before you press Publish for your review. 

[visualstudios]

11 minutes ago, donnovan86 said:

Well, one of my bad reviews is a person who clearly wrote good job and told me the work was great, yet he left 1 star. It was not a $5 order, it was a $50 order. Doesn't matter, he randomly left stars, which supports what I said above.

Multiple 4 star reviews are from people that did the same mistake you and others are saying regarding this new system, they thought they gave 5 and everything was fine. Again, randomly pressing buttons.

Multiple 3 star reviews saying: it's fine, he already published it, but he was expecting $100 worth of work in a $5-$10 article and stuff that was never included in the gig anyway. In the grand scheme of things it doesn't matter, but it's very frustrating to encounter these issues nonetheless. 

And all those together are what, less than 5% of your reviews? So it was rare, not "a lot of buyers", which is what you said. 1 in 20 is not "a lot". 

What we're seeing now is a much higher rate than that. That's the issue. If before 1 in 20 buyers left a random review, and now 1 in 5 (for example) leaves a random review, that means the system is much worse for accurate reviews.

Oh, and btw, of all the 1 star reviews you have, which are 33, you have 1 that was a mistake, the others are inline with the written review (which is certainly not random). So in that case, the "random pressing buttons" happened 1 in 33, or 3% of the time. Not "a lot of buyers" do it.

Finally, that 1 you got by mistake should be removed / corrected by Fiverr. It's unfair and a shame that Fiverr doesn't allow that to happen, even when buyers agree that it was a mistake.

Edited December 19, 2023 by visualstudios

[coerdelion]

41 minutes ago, edc_lab said:

Hello @catwriter thanks for precising that! 
I want to point out that I didn't use ChatGPT to write the core concept of my thoughts. However, this time, given the delicate nature of the topic, I asked to fix grammar errors to avoid silly mistakes.
Please keep in mind that gathering data, listening to feedback, and creating solutions around the end user are the basics of UX Design, I didn't write anything exceptional.
But again, thanks for the hint, perhaps you all appreciate more genuine mistakes over perfect texts (at least here on the forum, not in some articles you can find online ).

Your technical English is well above average.  Problem is, most people don't speak technical and *really* don't want to think much while they're reading. 

Reading age in most places other than technical / academic sites or papers should be around 14-15 years.  Makes it easy to read and understand, while passing the AI test. 

How do I know?  I put one or two of my essays from university through an AI detector. Was highly amused when they came back as AI generated. 

I graduated in 2008, well before AI was available.  And don't have a time machine

 

[catwriter]

50 minutes ago, edc_lab said:

But again, thanks for the hint, perhaps you all appreciate more genuine mistakes over perfect texts (at least here on the forum, not in some articles you can find online ).

As long as your English is understandable, it doesn't have to be perfect. 🙂 

And if you stay on the forum, you will see a plague of AI generated posts that bring nothing of value, and you'll (hopefully) understand the frustration some of us feel.

(We should probably spend less time on the forum, I guess)

59 minutes ago, edc_lab said:

Please keep in mind that gathering data, listening to feedback, and creating solutions around the end user are the basics of UX Design

That should be common sense in many things, not just UX Design. Unfortunately, common sense isn't all that common... 

12 minutes ago, coerdelion said:

I put one or two of my essays from university through an AI detector. Was highly amused when they came back as AI generated. 

Seriously? 🤣 I knew that AI detectors were flawed, but I didn't think they were that bad.

[coerdelion]

9 minutes ago, catwriter said:

As long as your English is understandable, it doesn't have to be perfect. 🙂 

And if you stay on the forum, you will see a plague of AI generated posts that bring nothing of value, and you'll (hopefully) understand the frustration some of us feel.

(We should probably spend less time on the forum, I guess)

That should be common sense in many things, not just UX Design. Unfortunately, common sense isn't all that common... 

Seriously? 🤣 I knew that AI detectors were flawed, but I didn't think they were that bad.

Lol! I got  a First - my essays were extremely robotic, full of relevant jargon and generally crafted in extremely "sophisticated" language. 

Not like I write here ...

[donnovan86]

51 minutes ago, visualstudios said:

And all those together are what, less than 5% of your reviews? So it was rare, not "a lot of buyers", which is what you said. 1 in 20 is not "a lot". 

 

1 in 20 is a lot, should be 0 out of 20. There shouldn't be any mistakes. But again, to my point, people randomly leave reviews. It's been an issue on Fiverr for a very long time. It's not something new and I don't attribute this to the new system. 

The fact that this new system shows even more so how random some people leave reviews doesn't surprise me. It's been like that for a long time and it will continue in that vein. Fiverr can't force people to pay attention to what review they leave. And they won't change reviews if they were left by mistake either, my point with that 1 star review I mentioned.

It is what it is, as far as I am concerned the new review system can be improved by removing the emojis and maybe redoing the 1-5 scale from very bad to very good, excellent does make people think twice and that might be the reason why some reviews are 4 or 4.5 stars with excellent text about the seller. Or again, people just rate randomly.

 

Again, buyers can see the review score so.. I don't understand where is a mistake being made, since they willingly pressed Publish for their review and saw their score before pressing that button. 

 

Edited December 19, 2023 by donnovan86

[donnovan86]

34 minutes ago, catwriter said:

Seriously? 🤣 I knew that AI detectors were flawed, but I didn't think they were that bad.

They are very bad. That's why if someone wants an AI check, I ask them what tool they trust and use that one. The reality is that I had the same text seen as both human and AI by different tools, so it's not something you can trust. And since it's AI searching for AI patterns, it's not like you can trust it. 

[emmaki]

AI detectors aren't great, in general:

https://arstechnica.com/information-technology/2023/07/why-ai-detectors-think-the-us-constitution-was-written-by-ai/

AI text is petty easy to see after a while - it really overuses a few words and phrases. Of course, you can tell it not to use those, but then it falls back on another level of overused stock words in the same phrases. It's the phrasing that's a dead giveaway.  If you chuck these texts into those AI tools that make AI content "undetectable," they mostly seem to work on making the grammar worse. Which makes sense, since another giveaway of AI is the high standard of grammar. 

A lot of students are getting into trouble at uni/college with Turnitin because it detects AI writing. I read a report somewhere where one student claimed Turnitin had marked their essay down to plagiarism - because they'd cited their references correctly. The problem is many of the professors etc. aren't aware that AI detection software is a bit rubbish. Of course, the students have plenty of incentive not to be entirely honest. And it seems the professors don't like dealing with the extra legwork of false-positive/reaching out to students etc.

The situation is comparable to Fiverr's use of AI to detect wrongdoing on the platform. Consider this: AI detects a gig has broken a copyright. The gig is deleted and cannot be restored. It later turns out that the AI misfired but "WELP! SORRY ABOUT THAT JUST MAKE A NEW GIG".

The new gig is a flop, with very little accountability. And now I'm wondering if AI's famous hallucinations might affect how these tools work, too. Yet this is how so many things are going: AI. Even normal people are passing on everything to AI. It wasn't so long ago that cellphones were only for the exceptionally rich: now (some) children get withdrawals if their iPad is taken away from them. It's not hard to see how the abuse of AI at every level could affect society in the future, I think. 

I am almost certain those little checkboxes in the emoji reviews feed into Fiverr's AI, and that is one of the key reasons behind the test. What else could explain the need for a 'value for money' metric, among the others? Remember, Neo needs data to match up with buyers. Take a look at any of the Neo promo videos. 

The emoiis? Just something for us to argue for 10+ pages about. Because Fiverr sellers love to moan 🙂

Still, I suppose we could thank AI for saving forums from the plague of "I'll just copy this other person's post, word for word and then claim I didn't copy it but I just liked it if I get caught". Whether the new plague is any better, well. 

I did write quite a bit more before, but it's in the red 🙂

[chronos_1]

9 hours ago, donnovan86 said:

1 in 20 is a lot, should be 0 out of 20. There shouldn't be any mistakes. But again, to my point, people randomly leave reviews. It's been an issue on Fiverr for a very long time. It's not something new and I don't attribute this to the new system

I think you are missing a point here. 1 in 20 is a lot and people are against it but the new system could make it 1 in 5/10, meaning it would be worse than before. It should be better than before about the review system

Other than that buyer do not know how the level system, rating system works here so according to them 4 out 5 star is great. Take Amazon for example, A product rating of 4 star is considered great. So the thing is either fiverr has to lower the overall bar for seller levels or create more education for buyers, I think first option is achievable. 

[catwriter]

12 hours ago, emmaki said:

And then they did the metrics thing so that if you don't do anything on Fiverr for ~60 days, your profile shows as 0 stars (this happened to me earlier this year lol) so that's already inbuilt and why you're not meant to vacay (I was not on vacation: I was sick. With lots of savings for this kind of thing, because you know, gig economy workers have no protections...) for more than 2 months... 

Were your gigs paused? And if not, how come you didn't lose a level? Wasn't there a warning in your analytics if there were no reviews in 60 days?

[emmaki]

Here's a post from Reddit on Fiverr's use of AI. If old sellers are leaving and/or scaling back due to Fiverr's issues and new sellers are getting this experience, what eventually happens? 

The amount of complaints from people on reddit complaining about experiencing this at signup initial gig creation has gone up a LOT in the past few months. This one is better than most due to the detail it provides. 

Meanwhile, at Fiverr....

"In our new campaign, we celebrate the unique talents, flaws, quirks, and emotional complexity of human beings, and explore the unlimited creative potential that’s unlocked when they’re partnered together with AI."

🤔 "Tell me about humans" indeed. "Imagine working with them" - we don't need to imagine Fiverr. We feel it, and it appears to be killing the morale of old sellers while killing the profiles of new sellers.

[emmaki]

3 minutes ago, catwriter said:

Were your gigs paused? And if not, how come you didn't lose a level? Wasn't there a warning in your analytics if there were no reviews in 60 days?

I was OOO. I did, of course, check in daily just in case I was messaged by old sellers. Which I loved. 

My metrics were at 100% throughout the whole time (they are always at 100%), but on my dashboard where all the metrics are, I noticed that my star rating dropped to 0. 

It went back to 5 when I got my first review on return. I'm guessing that while I was 0 on all my metrics due to having nothing for the metric system to eat on. Since no metric was under 100%, there was nothing for it to work with/act on, so my overall feedback went to 0 but my metrics stayed the same. 

A year or two ago I noticed that on my public profile, my overall review score was 4.9 and my dashboard was at 5. I contacted either CS or my SM about this discrepancy and was told that there was no problem and it was a bug in the system.

That bug has never been fixed. So anyway, dashboard score = your actual rolling feedback score and profile score = overall score. I guess?