app-mining
app-mining copied to clipboard
stacks-network
ProductHunt Analyzer
When I first came to AppMining, I didn't know ProductHunt very well, but I gave it a try, and it was an excellent experience. We met many new peoples to share our goal with them. But an honest question for me was, how others get these very high results? So I could try better next time. I did check them, and the result seemed sketchy to me. Then Blockstack added credibility votes, and it answered some of my questions. I thought PH guys did a rather good job there, but not enough. As a curious person, I wanted to check more these credibility of posts, so I gave it a day and faced interesting results.
What I did was, make an analyzer to get all the users who upvoted a post, and then check their other votes, creation time, followers, and more, from PH API. I looked for the similarity between two given users who did upvotes the same users and maybe if they created on the same day, as a meter to check if they are fake. It seems there are upvote sellers out there who made spam account mostly without or with dead Twitter accounts and stolen or fake identities. They use these accounts to upvote a product.
I attached the result tables, including a list of fake users and their properties, and also how many other users who voted for the same product, did vote just like that for different products. For example, two users, created on the same day, with no Twitter and upvoted only the same products. I calculated a similar score from SameVotedProducts/AllVotedProducts. Please note that for regular posts, this will not be the case, and you can see users will choose to upvote randomly; Not like bots upvotes just similar products in a same order and follow just the same people.
Here some sample results: Hint: The "PH Count" and "PH Analyzer Credible Count" are for last round, but "PH Credible Count" is from the provided excel file, so it is a month old, and they didn’t change much since then.
BitPatron: Link: https://www.producthunt.com/posts/bitpatron-3 PH Count: 1495 PH Credible Count: 454 PH Analyzer Credible Count: 170
Graphite: Link: https://www.producthunt.com/posts/graphite PH Count: 536 PH Credible Count: 497.3 PH Analyzer Credible Count: 520
BlackHole: Link: https://www.producthunt.com/posts/blackhole-file-transfer PH Count: 310 PH Credible Count: 198.8 PH Analyzer Credible Count: 301
Dmail: Link: https://www.producthunt.com/posts/dmail-3 PH Count: 1563 PH Credible Count: 1028.9 PH Analyzer Credible Count: 319
Here is a link to the tables: https://app.blackhole.run/#vDUgU9ig0L1FRzJnu4nor9Zw7JCVRgnzRqomsKp6nCkX
Some notes that may help you to check:
- I included a JSON and XLS version for you
- I like Firefox for opening the JSON File, it shows the results, and the links are clickable
- For checking the results, you can randomly pick any user from the list and open similar accounts, and you will get the idea.
- Use "Open Multiple URLs" Addon for opening a group of MatchedUsers https://chrome.google.com/webstore/detail/open-multiple-urls/oifijhaokejakekmnjmphonojcfkpbbh?hl=de
This is a one day work but seems accurate enough for now, but please let me know if you find any mistake so that I can look into it. My suggestion is that maybe we can work on this and use it as a credible vote of PH. Let me know if you have any questions.
As a sample out of the mining world, I just did a compare of two posts from yesterday. Figma Plugins and Snowball Money. Although (to this time of writing) Snowball gets 1270 votes, PH ranked it lower than Figma with 580 votes. Here is the detail: Figma Plugins: PH Count: 587 PH Analyzer Credible Count: 552
Snowball Money: PH Count: 1273 PH Analyzer Credible Count: 276
Snowball has a lot of votes form accounts like these: https://www.producthunt.com/@daisywcoward https://www.producthunt.com/@emiliestodd https://www.producthunt.com/@loisadaniels https://www.producthunt.com/@jillhbelisle https://www.producthunt.com/@rosemarygpeacock https://www.producthunt.com/@sheliagbraun https://www.producthunt.com/@lauracyoung https://www.producthunt.com/@tinacjohnson https://www.producthunt.com/@saratgeisler https://www.producthunt.com/@catherinevhewitt https://www.producthunt.com/@paulajnewton https://www.producthunt.com/@janetjwillingham https://www.producthunt.com/@wendykwinter https://www.producthunt.com/@leslieabarr https://www.producthunt.com/@lindarharris https://www.producthunt.com/@ednarbrowne https://www.producthunt.com/@rosakfavors https://www.producthunt.com/@judithhphillips https://www.producthunt.com/@mirandaglindstrom https://www.producthunt.com/@luellachiggins https://www.producthunt.com/@jeannercole https://www.producthunt.com/@leannesbarrios
The third place is Raising Venture Capital with 268 upvotes that 254 of them seems valid, so Analyzer result is similar to PH, the only difference is the count of the credible ones, and for that, I suggest checking the full report. Link to the full report: https://app.blackhole.run/#CCqCyortfX1CKZYqmWR5u8ezHa3wAoSswGf9a27qpi3a
Thank you @walterion. We're sharing this with the Product Hunt team.
Your welcome. Let me know if you need anything.
I think we 100% should look at this @GinaAbrams seems like @Walterion1 did great work here.
Great work @Walterion1 ! I have always been bearish on ProductHunt as an App reviewer...it is great to see some real hard data analytics and numbers.
Also - I think it would be appropriate that we run something internally just like we do with Awario to second audit PH. Frankly even Apple or Spotify launches rarely break 1000 upvotes.... should seem like a red flag.
I am not saying that we take on this work - but a simple audit like this would clearly signal gaming or not gaming. If we do take on this work - its very simple software like what was just built in this thread.
PH is a great tool if it is not gamed and the community on PH votes. It is a bad tool when we depend on PH to blindly audit for us.
This could have been identified with a random selection of upvotes. @GinaAbrams
Note from our contact at Product Hunt:
"From the looks of it there is a number of factors being left out here, it's not just about similarity and patterns. It's a pretty robust [system] and the factors listed by the author here don't account for the whole process behind how they are judged."
PBC discussed this. A few problems we see and potential solutions are described below:
-
The code would need to be open source. And would need to be deployed and maintained by a non App Mining entity. In our docs it already states that App Miners cannot be reviewers and this is just an extension of this. We want to avoid conflicts of interest.
-
The software would need to be adopted by PH. Or they would need to agree to use it in conjunction with their methods, then render final scores from this data. We want to avoid a situation where PBC or any App Mining entity adjudicates or modifies individual scores.
Really strange that they arent picking up on this...... There are countless examples of these accounts in the analysis. https://www.producthunt.com/@nanedela https://www.producthunt.com/@savetegy
Look at the profile pictures here :)
@GinaAbrams do you think this is a strange coincidence?
Or that there are users like this "voting" - https://www.producthunt.com/@hallermichelle https://www.producthunt.com/@hallermichelle1 https://www.producthunt.com/@hallermichelle2
It seems as though we need to do an audit just like we do with Awario - which is not tricky to handle internally.
Product hunt is a great tool if it is not gamed - PH is excellent if it is actually being used correctly.
"From the looks of it there is a number of factors being left out here, it's not just about similarity and patterns. It's a pretty robust [system] and the factors listed by the author here don't account for the whole process behind how they are judged."
I think that too, and I'm sure they have a much more sophisticated system behind there. The issue I was trying to point out is that the system could not find all these bots, and it leads to overwhelm of other apps launches.
The code would need to be open source. And would need to be deployed and maintained by a non App Mining entity. In our docs it already states that App Miners cannot be reviewers and this is just an extension of this. We want to avoid conflicts of interest.
I have no issue to share the code as I did before with the team, and I will be glad to help in anyway. Also I have no extra interest to put more work on myself but I will try to improve the community I have invest so much time on it.
The software would need to be adopted by PH. Or they would need to agree to use it in conjunction with their methods, then render final scores from this data. We want to avoid a situation where PBC or any App Mining entity adjudicates or modifies individual scores.
Personally, I don't think PH will adopt a new solution quickly as they want a system to cover all of their posts, and they have thousands. Here in Mining, we have much fewer apps, and the result can be audit much smoother. So I propose that PBC can add a layer of extra audition too it until PH may address this. Waiting for a longer time will extend the time of abusing the system.
PH may well have a very complex system in place that takes into account many features. That doesn't make it better. Clearly a relatively simple process is doing a better job than what they have in place. Is anyone arguing that PH's system is better?
+1 to the PHAnalyzer layer, with the addendum ... if PH does something to block the Analyzer from working in the future, they are dropped from app mining immediately, as a consequence.
Here is the result for Top20 apps for the last audit day (2019-08-15):
| App | Count | PH Credible | PHA Credible |
|---|---|---|---|
| Dmail | 1581 | 1065.8 | 310 |
| BentenSound | 669 | 626.6 | 665 |
| Arcane Docs | 408 | 375.6 | 390 |
| EZResize | 346 | 322.8 | 316 |
| Dadroit Viewer | 379 | 347.8 | 319 |
| BitPatron | 1503 | 498.7 | 185 |
| Arcane Photos | 273 | 265.3 | 254 |
| dPhone | 503 | 468.2 | 447 |
| Arcane Sheets | 278 | 264.9 | 266 |
| Lander | 525 | 446.7 | 427 |
| Bitcoin4Photos | 557 | 278.9 | 250 |
| Trove | 433 | 371.4 | 350 |
| Encrypt My Photos | 437 | 381.6 | 420 |
| Envelop | 247 | 219.1 | 231 |
| Graphite | 538 | 499.3 | 524 |
| BlockSurvey | 157 | 123.7 | 135 |
| BlackHole | 352 | 294.3 | 348 |
| Scannie | 167 | 139.1 | 145 |
| SpringRole | 568 | 467.4 | 414 |
| Forms.id | 435 | 397.2 | 421 |
Full report: https://app.blackhole.run/#45wS3BbwFm1LM851KuYscR7jrD9GmoXZhyRtLREhrACD
I hope we can do something reasonable about this issue as this seems growing: For example, Trove had 433 votes this past week, but like today it suddenly has around 900 upvotes, and exactly 360 votes are done in just one day (25th). The PHA Credible count is 440.
As an example, here are five persons with different names but the same avatars(!) and the equal votes:
- https://www.producthunt.com/@jaylin_marcus
- https://www.producthunt.com/@tedblake3
- https://www.producthunt.com/@luis_ca271
- https://www.producthunt.com/@byron_sabastian
- https://www.producthunt.com/@joey_mark1
The Blockstack apps in common with these accounts are:
- Trove
- SafeNotes
- Bitcoin4Photos
- BitPatron
- Pgeon
I honestly don't think Product Hunt's algorithm is nearly as good as the Blockstack team thinks it is. I have have seen plenty of evidence to support this. Product Hunt didn't need a strong algorithm early on like Reddit and Hacker News because it was a curated community. So they started behind the ball. Now, very simple voting schemes get past them.
Combine this with the fact that PH is the single most important part of App Mining, and something needs to change. I don't like the idea of removing app reviewers without a valid replacement ready to go, but if PH continues to carry the weight that it carries and continues to be something that can be bought, App Mining will see an influx of new apps, sure. But App Mining will also see a death of those apps within a few months when the money dries up.
Why are we not using the Audit tool the community has built, and agreed on, as the final score?
I think PH when not being gamed (hiring services / 3rd parties to mark upvotes), is a great tool.
Can we decide - like we did with Awario - to have an internal audit of PH, at least the top 20 / 30 scores.
Finally - Looking at the spread for PH Credible and PH Audit score, you can see clearly which apps might have to go through the audit process. There is no reason there should be a 600+ difference in these two scores.
PH algorithms are useful as they intended, hunting, and ranking. There were many posts before that bought votes, but no one cared much as what is the use of high votes? A tweet maybe! Here we use it as a Score, that is a door to let buyers buy the votes, and here is the spot we can check the results.
I value that Blockstack is willing to remove PH to show they are serious about the authenticity of the results. But I think removing it will dry the encouragement to make friendly and high-quality apps for the use of the outer and making it binary will probably have the same effect. Just Improving our current way faster to prevent the abuse will be a good start.
My propose is to use PHA on top of PH as a control layer to prevent this behavior. We are developers, and we can find a way for it, I think.
Product Hunt analyzer sounds like a good short/medium term improvement 👍 . Would need to be a 3rd party to run the software though right (think Blockstack PBC should remove itself and reduce complexity)? Someone who is not competitive in app mining to do it if Product Hunt doesn't accept it.. which may be the case as you can see from their response below:
Note from the contact at Product Hunt:
"From the looks of it there is a number of factors being left out here, it's not just about similarity and patterns. It's a pretty robust [system] and the factors listed by the author here don't account for the whole process behind how they are judged."
We can reach out to Product Hunt if we want to give them an ultimatum hat we indeed want to test this out for the next few months.
Related question: are the Product Hunt Analyzer rules gameable? For example, planting a garden of "fake users" that you can harvest months later. Are there others?
Also, I'm for removing them altogether and just having them be part of the onboarding since some apps don't launch well but improve a LOT over time. https://github.com/blockstack/app-mining/issues/144
There is a clear conflict of interests here, this is exactly opposite of what an Independent Reviewer should be. It is not different than an App Reviewer creating apps. I support removing, but not using an algorithm developed by an app miner. For example, if you asked me to do an algorithm, I would have a totally different approach, looking for problems in other apps. And this would happen with any app miner doing this. If this is used we should not call it Product Hunt Analyzer, because this is a Walterion Analyzer.
@pstan26 I see what you say; The reasoning about keeping PH was about promoting quality of ideas and apps as PH is the only one who controls that. Yes, any system can be gameable, either you keep it close by your self and go with it or make it as much hard as it was not worth the tray. Just like encryption algorithms, anyone can try to make all the values, but it only does not worth it.
About harvesting, I don't think any of these apps are doing this by them selfs, they just bought it with a meager price, and I believe harvesting votes will not worth the try as if they have time to do such task they would try to make better apps.
I updated the share code with the team. It is easy to read and change I think so PBC can handle it by itself for a while until we can find an alternative reviewer or PH do something by itself. I would be happy to help in any field if there was a need. Also, if you choose to go with this way, it can be a good idea to open it to the public to prevent any conflict or doubts, if it prevents such fake results, almost no one will have any problem except the fakers :)
This update was done concerning requests from some of the other miners. Actually removing PH will ease my job much as we have some apps, and controlling their launch is a heavy task.
Simply, whatever you do, please at least don't let some people get away very easily with hundreds of fake votes like past months. At least make it hard for them a bit ;)
I hoped that we see fewer gamers as we talk about the issue, but as it seems growing rapidly like an epidemy, I analyzed the issue again and here are some results.
Trove:
Sample list of 82 with very high similarity and most the voted apps got very high votes too and even the one that does not have may for randomizing purpose. It seems almost all of them are made in 1 day, they mostly followed by one man https://www.producthunt.com/@sachin_anchan and they seem fake as images are stolen from the web, take this as an example.
Exmape: Kendall Durgan https://www.producthunt.com/@kendalldurgan But here she is as "Tiffany Jenkins" https://jugglingthejenkins.com, she is a Sociologist and you can even search her name in Google! Example: Eliezer Ebert https://www.producthunt.com/@eliezerebert But here she is as "Jan Simpson" https://blog.texasbar.com/2011/04/articles/people/random-profile-jan-simpson-houston/ It seems these are stolen images from Facebook or somewhere and made with another name here.
https://www.producthunt.com/@yvonneromaguera https://www.producthunt.com/@beaulahsanford https://www.producthunt.com/@elysewehner https://www.producthunt.com/@tomasafriesen https://www.producthunt.com/@ginaemmerich https://www.producthunt.com/@katelinbuckridge https://www.producthunt.com/@richmondsatterfield https://www.producthunt.com/@clemensgutmann https://www.producthunt.com/@majornolan https://www.producthunt.com/@zoilablock https://www.producthunt.com/@idawillms https://www.producthunt.com/@lillieschultz https://www.producthunt.com/@kylarunte https://www.producthunt.com/@shawnfeeney1 https://www.producthunt.com/@hardywaters https://www.producthunt.com/@kenroob https://www.producthunt.com/@gretakihn https://www.producthunt.com/@eliezerebert https://www.producthunt.com/@jarretbalistreri https://www.producthunt.com/@nelsonweimann https://www.producthunt.com/@tyraschmitt https://www.producthunt.com/@lydiagleichner https://www.producthunt.com/@terrencecruickshank https://www.producthunt.com/@robbiebatz https://www.producthunt.com/@jadegraham https://www.producthunt.com/@matildehartmann https://www.producthunt.com/@casimerschulist https://www.producthunt.com/@jessiesmitham https://www.producthunt.com/@jazminauer https://www.producthunt.com/@aldaprohaska https://www.producthunt.com/@evertohara https://www.producthunt.com/@israelstokes https://www.producthunt.com/@margarettebradtke https://www.producthunt.com/@houstonmertz https://www.producthunt.com/@franciscarath https://www.producthunt.com/@donavonhermann https://www.producthunt.com/@genevievecronin https://www.producthunt.com/@kendalldurgan https://www.producthunt.com/@valeriebailey https://www.producthunt.com/@ricolind https://www.producthunt.com/@santinasteuber https://www.producthunt.com/@itzelmcglynn https://www.producthunt.com/@josecummings https://www.producthunt.com/@ilianalegros https://www.producthunt.com/@laverneroob https://www.producthunt.com/@jovannyjacobi https://www.producthunt.com/@deborahwest1 https://www.producthunt.com/@romandickens https://www.producthunt.com/@nelliehegmann https://www.producthunt.com/@randiosinski https://www.producthunt.com/@reillydibbert https://www.producthunt.com/@monroeglover https://www.producthunt.com/@laishabrakus https://www.producthunt.com/@wilhelmkovacek https://www.producthunt.com/@ovacormier https://www.producthunt.com/@ethylpacocha https://www.producthunt.com/@alessandrodeckow https://www.producthunt.com/@rebekasporer https://www.producthunt.com/@rahulokon https://www.producthunt.com/@dominickoss https://www.producthunt.com/@hermanmuller https://www.producthunt.com/@eldarice https://www.producthunt.com/@avisbraun https://www.producthunt.com/@sedrickbruen https://www.producthunt.com/@aliyahmorissette https://www.producthunt.com/@reidschmitt https://www.producthunt.com/@naomievon https://www.producthunt.com/@nikolasbeer https://www.producthunt.com/@hassanhahn https://www.producthunt.com/@kileyhauck https://www.producthunt.com/@sydniekutch https://www.producthunt.com/@jimmyschamberger https://www.producthunt.com/@laneypagac https://www.producthunt.com/@corinecrooks https://www.producthunt.com/@clayaltenwerth https://www.producthunt.com/@margarettekunze https://www.producthunt.com/@gerdametz https://www.producthunt.com/@felipaschmidt https://www.producthunt.com/@phyllisbauch https://www.producthunt.com/@albinaernser https://www.producthunt.com/@americabernier https://www.producthunt.com/@vitoleannon
Sample list of 97 with very high similarity, and it seems almost all of them are made in 1 day 2019-05-24 and almost 4 hours from 10 to 13 https://www.producthunt.com/@garfieldpaucek https://www.producthunt.com/@theresiahoeger https://www.producthunt.com/@tiahuel https://www.producthunt.com/@trentonlebsack https://www.producthunt.com/@sadyerice https://www.producthunt.com/@kevoncronin https://www.producthunt.com/@audiehirthe https://www.producthunt.com/@margietrantow https://www.producthunt.com/@jarvispfannerstill https://www.producthunt.com/@judyzboncak https://www.producthunt.com/@calehomenick https://www.producthunt.com/@kayleyroberts https://www.producthunt.com/@juliannelegros https://www.producthunt.com/@carolynschmitt https://www.producthunt.com/@terencehermiston https://www.producthunt.com/@venadouglas https://www.producthunt.com/@sandralebsack https://www.producthunt.com/@maddisondavis https://www.producthunt.com/@tonyschumm https://www.producthunt.com/@blairspencer https://www.producthunt.com/@aniyahmccullough https://www.producthunt.com/@louiegerhold https://www.producthunt.com/@mireillefahey https://www.producthunt.com/@manuelgleichner https://www.producthunt.com/@gailaufderhar https://www.producthunt.com/@paytonbuckridge https://www.producthunt.com/@quentinkovacek https://www.producthunt.com/@berniepurdy1 https://www.producthunt.com/@havenkessler https://www.producthunt.com/@ronaldojaskolski https://www.producthunt.com/@sabrinarenner https://www.producthunt.com/@gradyokuneva https://www.producthunt.com/@monserrateconroy https://www.producthunt.com/@craignitzsche https://www.producthunt.com/@stefaniestoltenberg https://www.producthunt.com/@itzeldibbert https://www.producthunt.com/@roelwunsch https://www.producthunt.com/@malcolmmosciski https://www.producthunt.com/@glorialeffler https://www.producthunt.com/@careyorn https://www.producthunt.com/@werneroconner https://www.producthunt.com/@kaylinquigley https://www.producthunt.com/@crawfordebert https://www.producthunt.com/@lorinegrady https://www.producthunt.com/@vernschimmel https://www.producthunt.com/@floriankuphal https://www.producthunt.com/@bartholomeoconnell https://www.producthunt.com/@albinabeahan https://www.producthunt.com/@daxnader https://www.producthunt.com/@rahulpacocha https://www.producthunt.com/@queeniestroman https://www.producthunt.com/@lilarosenbaum https://www.producthunt.com/@frederickjerde https://www.producthunt.com/@trecassin https://www.producthunt.com/@sisterboyer https://www.producthunt.com/@kurtisryan https://www.producthunt.com/@maevebashirian https://www.producthunt.com/@keenanebert https://www.producthunt.com/@hildegardrodriguez https://www.producthunt.com/@jabaribecker https://www.producthunt.com/@isabeloconner https://www.producthunt.com/@myajohnston https://www.producthunt.com/@brendanrunte https://www.producthunt.com/@vilmapowlowski https://www.producthunt.com/@orinwisoky https://www.producthunt.com/@jaedenhintz https://www.producthunt.com/@lindseywintheiser https://www.producthunt.com/@alexisorn https://www.producthunt.com/@elisecruickshank https://www.producthunt.com/@kobeleannon https://www.producthunt.com/@sonyakessler https://www.producthunt.com/@cordelljones1 https://www.producthunt.com/@petegutkowski https://www.producthunt.com/@tommiesenger https://www.producthunt.com/@abbigailbergnaum https://www.producthunt.com/@robynarmstrong1 https://www.producthunt.com/@dinawunsch https://www.producthunt.com/@rustyfritsch https://www.producthunt.com/@jaquandavis https://www.producthunt.com/@calistaleffler https://www.producthunt.com/@angiebrekke https://www.producthunt.com/@arjunborer https://www.producthunt.com/@lincolncassin https://www.producthunt.com/@johnathankuphal https://www.producthunt.com/@kennakuhlman https://www.producthunt.com/@hillaryhickle https://www.producthunt.com/@guycassin1 https://www.producthunt.com/@oswaldobosco https://www.producthunt.com/@cooperroob https://www.producthunt.com/@annamaedibbert https://www.producthunt.com/@audrabeatty https://www.producthunt.com/@kendallmosciski https://www.producthunt.com/@montanasanford https://www.producthunt.com/@justynbatz https://www.producthunt.com/@moniquehane https://www.producthunt.com/@allanmuller https://www.producthunt.com/@kiarracruickshank
There is almost 400 sample like these for Trove and much more for other posts like DMail, but I think you get my idea
Trove got 67 votes on the launch day and less than 20 votes next day and after that ten and even two a day and suddenly 360 votes on 2019-08-25 and just ten votes in a day after.
300 of that 360 users joined PH precisely three months ago in 2 days. They all voted for the exact same posts CookUp, Rivers IM, Real Dev, Trove and most of them get more than 1000 votes, and the rest are way less, and I suspect the owner put the for randomization, at least it is sain to do that.
2019-06-28 67 2019-06-29 16 2019-06-30 9 2019-07-01 8 2019-07-02 5 2019-07-03 2 2019-07-06 1 2019-07-08 2 2019-07-10 1 2019-07-12 1 2019-07-19 1 2019-08-01 1 2019-08-03 1 2019-08-04 2 2019-08-05 4 2019-08-06 45 2019-08-07 110 2019-08-08 83 2019-08-09 16 2019-08-10 10 2019-08-11 6 2019-08-12 19 2019-08-13 13 2019-08-14 10 2019-08-15 13 2019-08-16 11 2019-08-17 4 2019-08-18 8 2019-08-19 7 2019-08-20 7 2019-08-21 10 2019-08-22 8 2019-08-23 12 2019-08-24 4 2019-08-25 360 2019-08-26 10
A fun idea, check similarity between "Weights & Biases" (not in mining) and "Dmail". All have 1600 with sharing more than 1000 voters and with 73.3% similarity in voters and 45% with "Interactive Special Effects for Twitch" They share the same pattern to new apps like "Dapps.id", But it seems whoever controls the accounts, made new ones and now use them to randomize the results.
QRzebra was another one that is not in Blockstack. It got fake votes from exactly same people voted for Snowball Money, OptiMonk (not in mining), Dmail and more.
Dclouds:
Just like Trove, this one had a normal launch with 226 upvotes but now it has more than 1080.
Here is a sample list of 66 people joined before the launch on the same day 2019-09-25: https://www.producthunt.com/@ayu_mora_landy https://www.producthunt.com/@eirin_savira https://www.producthunt.com/@md_shofikul_islam https://www.producthunt.com/@teri_hredso https://www.producthunt.com/@ronny_denear https://www.producthunt.com/@sandro_sandoro https://www.producthunt.com/@mdsabuzhossain6 https://www.producthunt.com/@new_user_bff148e199 https://www.producthunt.com/@david_russo https://www.producthunt.com/@gaw_mic https://www.producthunt.com/@sudhir_sheoran1 https://www.producthunt.com/@md_fariduzaman_farid https://www.producthunt.com/@jose_lopez5 https://www.producthunt.com/@rakib_khan4 https://www.producthunt.com/@nenad_novi_juka https://www.producthunt.com/@akpos https://www.producthunt.com/@william_david https://www.producthunt.com/@jannatun_naher https://www.producthunt.com/@yanjmaa_batochir https://www.producthunt.com/@innovati_designs https://www.producthunt.com/@shiju_das https://www.producthunt.com/@md_rana3 https://www.producthunt.com/@bipul_bhattarai1 https://www.producthunt.com/@mehedi_hasan_bhuiyan https://www.producthunt.com/@song_world https://www.producthunt.com/@tapan_francis https://www.producthunt.com/@aasia_qasim1 https://www.producthunt.com/@james_jeremiah https://www.producthunt.com/@azizur_rahman1 https://www.producthunt.com/@theo_fillus https://www.producthunt.com/@sporsho_rejoan https://www.producthunt.com/@krishna_charan https://www.producthunt.com/@mubashshir_siddiqui https://www.producthunt.com/@mahammad_idrush https://www.producthunt.com/@sadhu_nakti https://www.producthunt.com/@hoanghieu1211 https://www.producthunt.com/@usama_tahir3 https://www.producthunt.com/@photoon_mael https://www.producthunt.com/@jonathan_dunlap1 https://www.producthunt.com/@rolando_m_kelly https://www.producthunt.com/@yosafat_beeny https://www.producthunt.com/@charles_kawel https://www.producthunt.com/@jasmine_zabila https://www.producthunt.com/@owl_of_nihility https://www.producthunt.com/@philip_cordeiro https://www.producthunt.com/@kezia_frans https://www.producthunt.com/@kevin_aming https://www.producthunt.com/@riaz_ahmed_baloch https://www.producthunt.com/@sumon_it https://www.producthunt.com/@borris_worrow https://www.producthunt.com/@anusha_paparaju1 https://www.producthunt.com/@fahad_ali2 https://www.producthunt.com/@manurali https://www.producthunt.com/@el_medi https://www.producthunt.com/@sh_sourav https://www.producthunt.com/@juju_bees https://www.producthunt.com/@lebu https://www.producthunt.com/@gopesh_chauhan https://www.producthunt.com/@dave_anand https://www.producthunt.com/@learning_english https://www.producthunt.com/@roselli_evana https://www.producthunt.com/@md_rajon https://www.producthunt.com/@kom_beshi https://www.producthunt.com/@mehedi93537384 https://www.producthunt.com/@shabiha_anjum https://www.producthunt.com/@abrahimboalia
Full report: https://app.blackhole.run/#leIgM7BioV1HhBExC5nkp4yfTaNUBGorWD7DaR5g2X79
MyPodium, SatsHi, dPhone and Lander
MyPodium and SatsHi are new, dPhone and Lander saw a +300 increase in upvotes from my last report. They are sharing between 300 to 500 voters with Trove (probably same seller) and even fake commenters like the case of MyPodium: https://www.producthunt.com/@hi_im_donglu https://www.producthunt.com/@richardkendri Almost all of the voters have no social linked to their PH account but these two have and if you check their twitter you an see there are exactly the same and tweets are almost identical, @cuevasm knows this behaviour from the Awario case.
Notes:
- They all getting all the votes after the first day as PH seems very sensitive to the first day (as it disrupts the ranking), as you check other normal apps you can see that highest votes are coming from the first day or two.
- These are my personal results, you should check them for your self, but it is probably too obvious to detect to normal eyes.
@pstan26 You are serious about PH situation, but gamers are still using the system, and in the meantime, it wastes our work. Is there anything we can do at least as a temp solution?
Here are another obvious 96 votes fro Dclouds. Can we at least do something about the obvious ones?
https://www.producthunt.com/@priyadcute https://www.producthunt.com/@ayu_mora_landy https://www.producthunt.com/@eirin_savira https://www.producthunt.com/@md_shofikul_islam https://www.producthunt.com/@teri_hredso https://www.producthunt.com/@mdsabuzhossain6 https://www.producthunt.com/@new_user_bff148e199 https://www.producthunt.com/@bhaskar_khadka https://www.producthunt.com/@nipaahmed8 https://www.producthunt.com/@ash_kumar https://www.producthunt.com/@david_russo https://www.producthunt.com/@gaw_mic https://www.producthunt.com/@alin_boicu https://www.producthunt.com/@sudhir_sheoran1 https://www.producthunt.com/@md_fariduzaman_farid https://www.producthunt.com/@rada_kitic https://www.producthunt.com/@jose_lopez5 https://www.producthunt.com/@rakib_khan4 https://www.producthunt.com/@aldijana_bratic1 https://www.producthunt.com/@nenad_novi_juka https://www.producthunt.com/@akpos https://www.producthunt.com/@william_david https://www.producthunt.com/@jannatun_naher https://www.producthunt.com/@i_sopadurso https://www.producthunt.com/@new_user_24889c8fbb https://www.producthunt.com/@yanjmaa_batochir https://www.producthunt.com/@innovati_designs https://www.producthunt.com/@shivaku35296536 https://www.producthunt.com/@ben_thomas1 https://www.producthunt.com/@bipul_bhattarai1 https://www.producthunt.com/@mehedi_hasan_bhuiyan https://www.producthunt.com/@song_world https://www.producthunt.com/@tapan_francis https://www.producthunt.com/@tiktok_bangla https://www.producthunt.com/@james_jeremiah https://www.producthunt.com/@theo_fillus https://www.producthunt.com/@sporsho_rejoan https://www.producthunt.com/@estherziitah_campo_s https://www.producthunt.com/@krishna_charan https://www.producthunt.com/@mubashshir_siddiqui https://www.producthunt.com/@mahammad_idrush https://www.producthunt.com/@sadhu_nakti https://www.producthunt.com/@hoanghieu1211 https://www.producthunt.com/@jessica007 https://www.producthunt.com/@usama_tahir3 https://www.producthunt.com/@photoon_mael https://www.producthunt.com/@jonathan_dunlap1 https://www.producthunt.com/@rolando_m_kelly https://www.producthunt.com/@yosafat_beeny https://www.producthunt.com/@charles_kawel https://www.producthunt.com/@jasmine_zabila https://www.producthunt.com/@owl_of_nihility https://www.producthunt.com/@philip_cordeiro https://www.producthunt.com/@kezia_frans https://www.producthunt.com/@kevin_aming https://www.producthunt.com/@riaz_ahmed_baloch https://www.producthunt.com/@new_user_2743aeca08 https://www.producthunt.com/@sumon_it https://www.producthunt.com/@borris_worrow https://www.producthunt.com/@anower_kabir_biswas https://www.producthunt.com/@anusha_paparaju1 https://www.producthunt.com/@navin_hedgar https://www.producthunt.com/@gilbertojrivasr https://www.producthunt.com/@fahad_ali2 https://www.producthunt.com/@rehema_mboka https://www.producthunt.com/@flyflybabe https://www.producthunt.com/@figueirafreddy https://www.producthunt.com/@marshal12502730 https://www.producthunt.com/@jorge_lara1010 https://www.producthunt.com/@manurali https://www.producthunt.com/@jesus_chafardet https://www.producthunt.com/@rays_pa https://www.producthunt.com/@el_medi https://www.producthunt.com/@pedrolp38815870 https://www.producthunt.com/@alexher922 https://www.producthunt.com/@asghar_abedi https://www.producthunt.com/@sh_sourav https://www.producthunt.com/@juju_bees https://www.producthunt.com/@lebu https://www.producthunt.com/@gopesh_chauhan https://www.producthunt.com/@kareena_sharma https://www.producthunt.com/@your_techtv https://www.producthunt.com/@gueder_ezer https://www.producthunt.com/@sanjay_varma1 https://www.producthunt.com/@maryam_khan3 https://www.producthunt.com/@learning_english https://www.producthunt.com/@roselli_evana https://www.producthunt.com/@md_rajon https://www.producthunt.com/@kom_beshi https://www.producthunt.com/@mehedi93537384 https://www.producthunt.com/@shabiha_anjum https://www.producthunt.com/@abrahimboalia https://www.producthunt.com/@josipjuric58 https://www.producthunt.com/@charity_muchoki https://www.producthunt.com/@sonam_sharma4 https://www.producthunt.com/@make_money2
@pstan26 A quicker solution will help the community at large. Delaying it will impact many honest players in the eco-system.
To get a solution up quickly, what about setting up an independent reviewer of PH marketing campaigns that for now primarily evaluates submissions from the community. This reviewer could accept arguments and analysis from different parties to avoid being one-sided.
This approach could encourage rapid development of different approaches to analyze and control for unfair practices. See also the related issue #155 on adding a reviewer for marketing practices.
@GinaAbrams a solution to that I remember Facebook, Twitter, and Telegram has used before and probably PH should use is to double check IP addresses of the people who are voting, making sure there isn't a Product Hunt farm somewhere.
For example just by doing some Google search I found this website who lets you buy 400 PH votes for $270. PH should buy their services and choose a fake app to detect all fake accounts and IPs and block it.
https://www.appsally.com/products/product-hunt-upvotes/
Also, make it clear to App Miners that if it was discovered that you used a paid PH service, your app will or you will be banned from app mining for 3-6 months
Once again. We've been tracking this issue for 6 months in git, and a YEAR in the forum. Time to call over. There is no more time for experimentation. I highly encourage everyone to launch on PH, but they should no longer be a reviewer.
As an example that it is still growing here are more samples:
Magic Spoon
It is only one day passed and it has near 400 votes and PHA credible vote count is only 100. Users like these and the difference you see is because the one who makes these accounts use these random upvotes to mislead the PH and/or make things seem normal. There share votes for some of the ones I reported before (not in Mining). Like River IM, Hexometer and Gooroo.
All the accounts seem fake and images are stolen just like the other like Trove case. Example: This account image is stolen from this resource. https://www.producthunt.com/@vincelindgren https://blog.texasbar.com/2013/04/articles/people/random-profile-mary-grace-ruden-houston/ Example: Users with the same images and different names and similar votes: https://www.producthunt.com/@lamersnovella https://www.producthunt.com/@couzensclaudia
Here are some samples: https://www.producthunt.com/@bernicebarris https://www.producthunt.com/@corazoncrozier https://www.producthunt.com/@argeliaauxier https://www.producthunt.com/@chantellcostin https://www.producthunt.com/@shaunaspradling https://www.producthunt.com/@cathicendejas https://www.producthunt.com/@marybellemalmberg https://www.producthunt.com/@dannielledeloera https://www.producthunt.com/@simonescheetz https://www.producthunt.com/@tonietrojan https://www.producthunt.com/@apoloniaallgood https://www.producthunt.com/@martinamullenax https://www.producthunt.com/@paulettapastrana https://www.producthunt.com/@herlindahoney https://www.producthunt.com/@athenatorp https://www.producthunt.com/@reedwest https://www.producthunt.com/@vincelindgren https://www.producthunt.com/@estefaniadurgan https://www.producthunt.com/@verniebrekke https://www.producthunt.com/@namekuhic
Codelets
It is the same for other new apps like Codelets. It has only 60 credible votes in its 260 upvotes by now, it means less than 25% credibility in votes. Take into account that the other apps from the same developer, Lists have 170 credible of 353 votes. Keep into account that many of these voters are sharing votes with other noted apps in the Mining like dClouds.
Note: For remembering how a real voted apps look likes, Recall has 430 of 443, that means 97% credibility in votes! It is almost the same for apps like Graphite, BlackHole and Scannie.
Full report: https://app.blackhole.run/#8BstWEvAWE1BhKFcBXkYFnnewEeTLD777jLKp9BMRFPA
Conclusion:
So I think keeping PH can be excellent just like Awario, but we should have something like Blened Awareness to clean the votes. Removing it will not remove the effect that these apps had on Mining for months, and discourage new app developers from competing in making and launching hot apps that make more attention to Blockstack and community. I know PH didn't provide more info on their internal system, but there is no need to ask them for that, we can use a tool or reviewer to audit. It is just like the Awareness, we don't ask Twitter or Instagram for real data, we ask Awario to gather it, and we tried to have a good way to audit them. And for doing that we don't even need PHA to worry about conflict of interest. We can have a short term way as @pstan26 stated to clear the Mining from fake scores and work on improving the review process in the meantime.
Please keep in mind that I don't know any of the apps or developers, and I don't know what their reasons to do such are. As a developer, I found a bug, and I propose a fix to it and not removing the feature.
Hope this helps the community.
A high-level comment here that app miners pointing out potential issues/weakness in app reviewers by using public data is a healthy sign.
We should not assume that the reviewers are perfect; we should help them improve.
I do see potential conflicts of interest here as well and might make sense to take our time to carefully address any issues. The idea of an independent team "cleaning up" scores is interesting but we should think about how will that be sustainable in the long-term.
I think HN a done a lot of work on how upvotes work and there might be lessons that can be applied. (We can even talk to the HN team about it.)
@muneeb-ali thanks, I hope we can find a better way to address this issue. As a quick update: In this weekend it seems Codelets bought near 400 and (currently has 630) new upvotes and BentenSound bought near 600 new upvotes (currently has 1200). A fast-growing epidemy that probably we will find more as such this month, and we cant remove PH this month as it will be unfair to people (including my team) that tried to have a powerful launch in a fairway.