This is not a good take in my opinion for the moderation part. I see no fun what so ever in suspending over like 30 accounts in a week. It’s terrible and overwhelming. Probably over 100 suspensions have occurred in just Macao in the past week. i feel like I’ve gone from a volunteer to a staff member with the amount of time and effort I’m putting in moderating.
Suspensions should never be fun for any party involved.
and it especially strikes me that there’s little chance to “emerge from this stronger” if nothing changes. if the only point with the CNC is to do more of everything — reach out to more students, expand to more regions, make more observations — the number of sockpuppets, data fraud, and so on will only grow. never mind the trend being unsustainable — things aee borderline insurmountable to clean up even as they are right now.
If anything we are weaker as its likely some people might leave inaturalist becuase of these events the problems created and lack of good solutions.
If anything, this shows how inaturalist needs many reforms to better scale with growth. I dont think its a coincidence after the CNC, tools and guidelines for moderation are a “top priority” says Tony (iNat staff).
CNC observers need to learn - like a puppy is for life, not for Christmas.
Stick around. Come back. At least notice, then withdraw your wrong IDs so we are not fighting the Community Taxon algorithm, on your behalf.
75% of IDs by 1% of identifiers. iNat needs to carrot and stick the 99% to do … a little MORE!
At midnight I will turn into a pumpkin. Or a glass slipper …
@dianastuder, I definitely feel like a squash after almost 10 days of the City Nature Challenge, so we can be pumpkins together.
Even though I haven’t run into really awful problem observations for the CNC, I do feel overwhelmed by the sheer volume of perfectly good observations. It’s really spring here in the northeastern United States and all forms of life, including iNat observers, are waking up.
FWIW the timing is more about being able to focus on implementing medium-to-large-scale changes after CNC rather than making those large change soon beforehand. Improving the way suspensions work has been on my mind for a while. While I agree the issues with CNC highlight the need and I’m frustrated by the way things are currently working during it, in this case it was going to be a priority project of mine after this CNC regardless of what happened. Correlation isn’t always causation, so please don’t jump to conclusions.
I’m glad that’s the case, as I wouldn’t want to rely on a significant disruption to the system to be the only impetus to make improvements. Perhaps that’s another reason I’m leery of the framing in the OP of this thread. We would need the coming changes regardless of how the CNC went.
I can clearly see and appreciate the overwhelm and frustration for moderators and dedicated identifiers. Please just know that there are those of us that see your efforts, and THANK YOU.
It is frustrating that the numbers seem more important than quality, that so many people just have some ‘fun’ for a weekend and then forget about it while others have to clean up. And dishonesty - how very disappointing.
Unfortunately iNat is a slice of life itself, people will be people, with all the good and bad. And when it concerns nature, I also often feel that for every step forward, miles are lost. But I can only do my best with the steps forward, and for my own mental health, I can choose to focus on the new observers that are hooked, the special observations, the dedication of the volunteers and staff, and of course, my own joy at participating.
Thanks to every one of you who ID, moderate, do the dirty work while we are having fun.
Encouraging number for Africa and CNC
https://www.inaturalist.org/projects/225199/journal/111171
I don’t know where all this “bad behaviour” took place but it certainly was not eveywhere. I have never really noticed anything like this. Neither during my normal use of iNat nor during the CNC period. I may be blissfully ignorant but in this case ignorance is bliss I think. inat is so full of good people, good fun and fabulous encounters with nature that these mishaps must be a small minority and by all well meaning iNat lovers should rather be ignored than get frustrated about. Using iNat may seem like a game in some ways but there are no winners or losers so both cheating to win as well as the envy of not winning are totally obsolete and completely defy the point of having fun, enjoying nature and interacting with like-minded people
I’m glad that you, and apparently several others who have posted in this thread, did not have to see the large volumes of cheating observations and rules-breaking behaviour that took place during the CNC. It’s by all means better for your sanity and peace of mind not to see all the negative things that came with it.
On the other hand, the fewer people there are who see and deal with the problems, the larger the volume of problems there will be for each person to try to fix. Unfortunately, for those of us who use and value iNaturalist as an educational or scientific resource, “ignoring rather than getting frustrated about” the massive influx of bad-faith content is not an option in the long run. Fake observations or out-of-place organism records really do degrade cases like – just to name a couple things I’ve used iNaturalist for – teaching biogeography and downloading range data. Not to mention, future updates of iNaturalist’s own Computer Vision and GeoModel suggestions depend on the quality of information going in, and incorrect or fraudulent posts will damage the site for the rest of us seeking to learn from those suggestions.
Obviously not everyone can, or needs to, trawl through hundreds of iNaturalist observations looking for copyright violations and spoofed locations. I encourage everyone to do their part, and to address anything problematic that comes across your dashboard, but it’s perfectly okay if you don’t have the time or will to specifically seek out false accounts or other violations for the depressing task of mass flagging. Even so, someone has to do it eventually, and it’s all volunteer work, which will inevitably fall to someone else who’s also just trying to enjoy nature through iNaturalist.
For me, the major frustrations with the CNC weren’t so much outright malicious activity, but rather the floods of observations and CV misidentifications in taxa where there’s only a handful of trustworthy identifiers. Trying to keep up with and correct “CV take the wheel” IDs in terms of both initial IDs and people either blind-agreeing or semi-randomly picking CV suggestions has absolutely burned me out over the past couple of weeks.
You can use the Geomodel Anomaly to find them.
But that grunt work needs to be done by Not the taxon specialists.
I was browsing the geomodel anomaly records a while ago, and wondering why there were only common species in the list. Then I remembered that misidentifications of species rare enough to not be covered by the Computer Vision are also not covered by the geomodel…
The Geomodel is the most concerning thing. If even a handful of American, Australian, whatever, get two confirming IDs or more. It very well could influence the model depending on how much data the taxon has.
This means American, Australian, etc species could start getting reccomended in places they shouldn’t. Which means a 100% CV inaccuracy rate for the taxon in question it suggests there. Ive been fighting a Nephrotoma species that was blindly agreed to in Taiwan. About 8 RG obs were enough, and now every week, maybe 40 are uploaded around asia. A continent it doesn’t live in.
This bad data has real consequences. This is why i have hope not only will sockpuppet IDs be removed, but the IDs of accounts that mass blindly agreed to all the observations of another person. While harder and more tricky to diagnose then sockpuppets, becuase of IPs and other evidence that sockpuppets can have. The actions of some people IDing just their friends observations or vice versa has the same outcome as if they are a sockpuppet.
It is the difference between doing it myself and creating a new account to agree to my obs, or asking a friend for a favor and doing it for me.
I think iNat suffers from a scalability problem. It’s easier to observe interesting organisms than it is to learn to ID them reliably, even assuming good will.
I don’t want to get too negative about the City Nature Challenge, having said my piece quite a few times already. I do want to emphasise that I think all the organising of it is done with very good intentions. Still, as in the saying, we don’t need a neon signpost to know where the road paved with good intentions leads.
Without a doubt, this is true. It always has had the problem and it always will. A toxic dump of millions of observations via the CNC is the apotheosis of the issue. To the extent that the new accounts created for the event are leaving any identificiations, they are most often actively bad IDs that are in pursuit of making observations Research Grade with no interest in quality. Obviously I want more observations and data; rarely seen powdery mildews or remote mountain flowers being my stock-in-trade, and they’re uploaded at a low rate, so I would love encouraging more people with crop woes or living in high-altitude villages to take more photos. The CNC on the other hand is clearly “for the kids” figuratively speaking, and the data quality is an afterthought if it’s any sort of thought at all. I don’t think even researchers and educators looking at urban plants and wildlife will be able to get anything useful out of the data as it currently stands. Give it a year and perhaps we cleanup volunteers will get it 80% fixed! Just in time for what I’m sure will be billed as the “biggest CNC ever”. Just in time for frauds in large metropolitan areas to open the floodgates with more stolen images, mislabeled observations, and captive organisms, to boost their city to the top of the charts once again. It’s only a very specific subset of users who will ever cheat, and I think even a very specific subset of cities that will see the cheating happen – I’m happy for those who can ignore the problem – but that subset will be back.
I didn’t realise it would be as few as the half-dozen or so that you mention. All the more reason that I tend to be more rather than less aggressive when bumping the community taxon ID or marking a questionable plant or animal as captive/cultivated. I’ve honestly been uncertain of whether the thresholds of observations/photos for incorporation into the Geomodel / Computer Vision are high enough – though I do feel confident that that’s at least something the iNat staff weighed very carefully, the Geomodel updates seem to be more easily influenced by small amounts of data than do the Computer Vision updates.
This is correct regarding the geomodel. I have put heavy emphasis on curating it for Chironomids. With general knowledge on the system, it actually can be easy to manipulate and change, *easiness depending on context of taxon. Camptochironomus, Glyptotendipes, Diamesinae, are some that i have put work into curating for the geomodel specifically. Diamesinae like mountains and extreme latitudes. I have been quite successful getting the geomodel to reflect that.
Other details of other taxa not really relevant. The point is one can create a plan and purposely ID in a way to promote a stronger more correct geomodel… or the other way around making it worse.
I really do worry about this becuase its not as simple as. “Oh some people blindly agreed on 1000s of observations.” “Ok, we’ll get them fixed eventually they aren’t a big deal and are a small percentage”
No. These misidentifications can create more misidentifications if conditions are right creating a feedback loop if not taken care of. So it is a concern.
I check Geomodel Anomalies for the Cape Peninsula every day.
Is it an Anomaly, or a local Endemic in an area too small for the Geomodel’s blocks?
(The wording and function has changed a little, but it is in my head as it was) CV suggests Seen Nearby. I go to the map hunt down That One - it used to be, a single obs, with a single ID - and then it spiralled, if identifiers didn’t fight back. If we can clear the older problems, there is less going forward. Yesterday it was Typha capensis (yes) versus dominguensis (no, not in our South).
I leave a comment - since I have tracked down the info anyway
You have chosen a sp from … tweaked to fit each obs.
And the other way. If I am checking distribution maps and I don’t expect to see The World with Weird Outliers. While remembering invasives and garden plants, sigh … once burnt, now very wary.
You have chosen a sp from South Africa. Might be … definitely dicot … whatever
Multiple people (perhaps on different threads) expressed interest on how the CNC results would really shake out if given more time for “clean up.” To explore that question, here’s a comparison of the official results 2023 (two years ago) vs what that umbrella project is still reporting. I am omitting the “Global Project” from the rankings, since it is on the umbrella project but not on the CNC website.
The biggest change, as one might expect, is in the ranking of teams by number of species. For the most part, the top 10 are still in the top 10, with two major exceptions.
| City | Species Official Result | Rank | City | Species Current Reporting |
|---|---|---|---|---|
| La Paz, Región Metropolitana de La Paz, Bolivia | 5,344 | 1 | Hong Kong | 3,718 |
| Hong Kong SAR, Hong Kong SAR, China | 4,469 | 2 | City of Cape Town | 3,553 |
| Cosalá, Sinaloa, Mexico | 3,912 | 3 | La Paz (CNC) | 3,339 |
| City of Cape Town, Western Cape, South Africa | 3,847 | 4 | Graz bis Tierwelt Herberstein | 3,303 |
| Houston-Galveston, Texas, United States of America | 3,707 | 5 | Houston-Galveston | 3,241 |
| Graz bis Tierpark Herberstein, Styria, Austria | 3,688 | 6 | South Florida (#CNCSOFLO) | 2,780 |
| South Florida (#SOFLO), Florida, United States of America | 3,094 | 7 | Chiayi-Yunlin | 2,617 |
| Dallas/Fort Worth, Texas, United States of America | 3,065 | 8 | Dallas/Fort Worth | 2,606 |
| Washington DC Metropolitan Area, DC/VA/MD/WV, United States of America | 2,931 | 9 | Washington DC Metro Area | 2,550 |
| San Francisco Bay Area, California, United States of America | 2,852 | 10 | San Francisco Bay Area | 2,401 |
Where did Cosalá go? They actually fell from third place all the way to 19th place, now reporting only 2,014 species. Chiayi-Yunlin rose up to 7th place from a previous 14th place.
The rankings for number of observations and number of observers haven’t changed too much. Some times the numbers have gone up slightly (I assume some participants missed the upload deadline) and sometimes they go down (I assume this would indicate some observations/accounts have been deleted/suspended).
| City | Observations Official Result | Rank | City | Observations Current Reporting |
|---|---|---|---|---|
| La Paz, Región Metropolitana de La Paz, Bolivia | 126,435 | 1 | La Paz (CNC) | 124,458 |
| City of Cape Town, Western Cape, South Africa | 52,518 | 2 | City of Cape Town | 52,300 |
| Dallas/Fort Worth, Texas, United States of America | 48,021 | 3 | Dallas/Fort Worth | 48,655 |
| Mazatlán, Sinaloa, Mexico | 42,479 | 4 | Houston-Galveston | 42,438 |
| Houston-Galveston, Texas, United States of America | 41,736 | 5 | Mazatlán, Sinaloa | 42,165 |
| Hong Kong SAR, Hong Kong SAR, China | 39,071 | 6 | Hong Kong | 39,367 |
| Washington DC Metropolitan Area, DC/VA/MD/WV, United States of America | 37,503 | 7 | Washington DC Metro Area | 37,984 |
| Cosalá, Sinaloa, Mexico | 36,644 | 8 | Cosalá, Sinaloa | 35,802 |
| Graz bis Tierpark Herberstein, Styria, Austria | 33,106 | 9 | San Francisco Bay Area | 32,979 |
| Monterrey Zona Metropolitana, Nuevo Leon, Mexico | 32,680 | 10 | Graz bis Tierwelt Herberstein | 32,883 |
| City | Observers Official Result | Rank | City | Observers Current Reporting |
|---|---|---|---|---|
| La Paz, Región Metropolitana de La Paz, Bolivia | 3,025 | 1 | La Paz (CNC) | 3,011 |
| San Francisco Bay Area, California, United States of America | 2,488 | 2 | San Francisco Bay Area | 2,746 |
| Los Angeles County, California, United States of America | 1,671 | 3 | Los Angeles County | 1,826 |
| Monterrey Zona Metropolitana, Nuevo Leon, Mexico | 1,655 | 4 | Washington DC Metro Area | 1,626 |
| Washington DC Metropolitan Area, DC/VA/MD/WV, United States of America | 1,527 | 5 | Monterrey Zona Metropolitana | 1,613 |
| Houston-Galveston, Texas, United States of America | 1,477 | 6 | Houston-Galveston | 1,571 |
| Boston Area, Massachusetts, United States of America | 1,456 | 7 | Boston Area | 1,545 |
| Dallas/Fort Worth, Texas, United States of America | 1,305 | 8 | Dallas/Fort Worth | 1,385 |
| City of Cape Town, Western Cape, South Africa | 1,284 | 9 | Greater Austin | 1,303 |
| Greater Austin, Texas, United States of America | 1,180 | 10 | City of Cape Town | 1,296 |