The OpenScience Laboratory The Open University

iSpot Forum

Re iSpot changes - from the iSpot Team

Thank you so much Tony - done it, and it works. Great that the thumbnail view is what you’ve asked for, I would imagine that if it’s the default option it would be great for many/most of the ‘citizen science’ users.
Thanks again,
Tessa

Quicker:
MySPot > Observations - ignore the useless page and click on MAP.

This opens your LIST/GALLERY/MAP set of pages like your old iSpot MySpot > Observations.
(except if you look you will see that this is as DavidHowdon said - just a filter with your own USER. Except that you dont have to go through the filter steps.

You can also save it in your Saved Filters.

The main issue I have with iNaturalist (which I have used on occasion) is that there is no way of telling whether a person providing the identification has any competence or skill. At least I cannot find one, although I’ve only used your iOS app so it may be it is more obvious on web browser based system.

iSpot’s reputation system is far from perfect but does at least try to give you some insight into whether or not you should regard the support provided as authoritative. Unless iNaturalist has something like that I doubt I would use it in preference to iSpot.

I also know what iSpot was set up there was a lot of pump priming with the team going around natural history societies and encouraging us to get involved and to get our ‘expert’ members badged . I know that’s how I got involved. iNaturalist seems quite light on people who can help with IDs (here in the UK at least) which makes it quite slow in getting more that cursory responses and limited in providing the sort of support in developing ID skills that iSpot was designed to provide.

1 Like

Yes - the Reputation System is clearly a huge strength of iSpot, I’ve read Jonathan Silverton’s scientific articles on it and its very clever. We’re in the midst of trying to build a reputation system on iNat and I’d love to learn more about what works/what doesn’t work on iSpot. We’re currently running an ‘Identification Quality Experiment’ where we’re registering and collecting blind IDs from taxonomic experts: http://www.naturalista.mx/pages/identification_quality_experiment (if you can help, please register!)
we’re using these data to try to quantify how accurate IDs are on iNat and use these to come up with a reputation system for weighting IDs, you can read an update here on the progress of this experiment here: http://www.naturalista.mx/journal/loarie/10016-identification-quality-experiment-update

If you could rebuild the iSpot Identification system is there anything in particular you’d change? Or does it work pretty well as is? My understanding is that the categories are pretty coarse right, ie you earn expertise in ‘Plants’ or ‘Mammals’ we were thinking of tracking and awarding expertise on a per species basis, ie so-and-so has earned reputation X with species Y. Do you think that would work?

Update: @dejayM and @AlexanderR thanks very much for the input. I’ve also responded to each of you but my original post and those responses were flagged as spam by the developers. Not sure why - Would love to continue the conversation…

1 Like

Plants (for example) is too coarse - the usual specialisations are seed plants, ferns and horsetails, mosses and liverworts, and algae (which aren’t a clade), with club mosses, hornworts and stoneworts sitting uncomfortably alone. But make the categories too narrow and it takes too long to built a reputation.

I don’t know whether saturation (every long term user ending up with the same effective weight) is a problem with iSpot’s
reputation system. Tony may know.

Giving out Specialist (new), Expert and Knowledgeable badges based on narrow taxonomic groups and (restricted geographical regions) would work - a British fern expert could be marked as Expert, while being treated as Knowledgeable or even one of the hoi polloi for seed plants.

PS: your previous messages may have been flagged as inappropriate rather than spam. I wouldn’t rush to the conclusion that it’s the developers that flagged them.

2 Likes

I have a schema for reputation based on the taxonomic tree. it is quite simple and elegant, and not too computationally intense. it is not tied to any group. I would avoid species. Start at the genus level. You need to sum reputation based on number of IDs, and to simplify the schema you need to average them over species. The problem is the huge number of nodes if you work at species level: there are more than enough on the taxonomic tree. Reputation is constant down a clade, but decays upwards, allowing an expertise in Scarab Beetles to carry much weight with other beetles, and less with Bugs, and much less with non-insects and almost none across to vertebrates. Users can independently acquire reputation in any taxon, but superior users would slowly acquire reputation at higher and higher levels. Recognised taxonomists would get expert status in their specific taxa.
Unfortunately, iSpot is far from ready for such a schema.

2 Likes

I’ve been rethinking reputation lately. Why does reputation continue to increase over time as you post IDs? If you use the site for 10 years you will easily reach max rep, but it doesn’t mean you have increased your skills. Can iSpot not rather work with a more sophisticated algorithm that incorporates the proportion of likely IDs for the users last 100 or 1000 IDs?

This could then be linked to your taxon framework.

1 Like

One can do anything.
But there are computer overheads: dont make computation too difficult.
And there are social consequences: this site includes adults and kids - so rewarding is important, and not punishing (lower scores) is also important.
But think about it. Learning how to identify is a skill learned that is not easy to forget. The purpose of iSpot is to get users to learn to ID, which is a mentor-student relationship. Once learned the skill persists. The aim is not to get people to rote learn differences. The aim is not to have beginners knowing by site all 900 bird species. iSpot is an ID tool as well, and learning to use iSpot (unless the programmers mess it up) is a learning curve, that takes mere hours to refresh even after several months or years absence.
So the real question is simply. How many IDs does one have to make to become proficient in a group? The answer is probably in the region of 200 to 1000 IDs. (for iSpot it is 500 - a silver or knowledgeable user has 500 reputation, and to get to this level, with a maximum of one reputation per ID, requires 500 correct IDs agreed by an expert, or two knowledgeables. If agreements are only by novices more IDs are needed!) Now consider - there are dozens of proficient users in iSpot in any group, so a novice entering the system will probably have to post their own IDs, so this is a very tall ask. There is a flaw on iSpot, but I have not yet seen any users exploit it mercilessly: one can post 500 observations of a common, easy to ID, species and earn that reputation. But the amount of cheating has not warranted adding a far more complex algorithm to cap or filter reputation.
Anyone who has worked on iSpot for 10 years is probably highly proficient in their group.

A bigger issue is geographical expertise. iSpot does not recognize that an expert in African birds may be a novice in Asian or South American birds. But the logic does apply: if I am proficient at identifying African birds, I should easily be able to apply my skills to American birds.

Note that iSpot reputation has nothing to do with being able to make an ID spontaneously without reference to iSpot/books/keys/other IDs tools. Reputation on iSpot is not about knowing the species, but being able to ID them. Spontaneous ID of 900 taxa in your field of interest is an entirely different ballgame: and your ideas above would apply. But memorizing 900 taxa is not necessary on iSpot - there are multiple examples of all species on iSpot - iSpot is the collective memory and the user just needs to learn how to tap into that - with the help of resident experts and more experienced users.
There is a need on iSpot though to recognize more invertebrate groups: skills in insects are not quite the same as for arachnids (or at least they are far more different than the gap between the vertebrate groups).

3 Likes

Can we have a list of planned features (and estimated timescales) please.

1 Like

Yes, I took a look too. Wondering where best to invest my time and effort. Think I’ll hold fire for a while and see how things pan out …
.
Still think we need a proper community-run platform
.
Along the lines of a P2P Foundation vision:
.
https://blog.p2pfoundation.net/from-platform-cooperativism-to-protocol-cooperativism/2017/07/05/comment-page-1#comment-1578866

Reiterating @lavateraguy’s comment
"Can we have a list of planned features (and estimated timescales) please."?
PLEASE

1 Like

This build on the new platform was scheduled for October. So my guess is that everything that is going to be done will be done by then. A bit like Sept 2014 when the teams departed and we were left with the mess. Of course, we will only get a mess if we allow it.
But changes is an add on. It was not included in the original scoping. The question is whether the development team with spend the time until October doing CHANGES instead of debugging (Changes is more important than many of the bugs) or if the debugging will be added onto the time taken from CHANGES. We will see.
It is not usual for iSpot to tell us the schedules. And if you really press them, that is because they usually dont know. The programmers are a law unto themselves and us mere iSpot mortals must quiver and take what scraps they deign to discard for our pleasure. Consult with us? Tell us anything? You must be mad!

You are more sane than I thought! But, I suggest, some of the bugs, deficiencies and tweaks will be easier to fix.

The categories are pretty course, but that is a deliberate choice and I understand that thought has gone into it. As I was originally recruited to this site iSpot was designed to be the ‘friendly uncle’ who takes a budding naturalist out and shows them things. So necessarily the categories were set at quite a high level - if you are just beginning you might know the difference between a fly (diptera) and caddisfly (trichoptera) so if before you could contribute you had to work that out it would be off putting. Actually I get the impression (subjectively rather than based on real evidence) that even the current ‘invertebrate’ category causes confusion with a lot of new people putting stuff down as ‘other organism’ rather than one of then inverty-thingummies.

A similar reasoning was behind the choice of only a limited number of habitat types.

Early on I suggested that the iSpot reputation system should allow you to lose reputation for getting stuff wrong as well as gaining it for getting it right. In the real world that is how it works, both getting it right and not getting it wrong matters. iSpot deliberately chose not to do this as they had a philosophy of nothing negative happening to people (again to encourage people just venturing into natural history observation) and losing reputation points would count as negative. That argument has a lot of merit.

The major problem I have with the habitats is that approaching 50% of my observations - mostly various classes of urban ruderals - don’t fit any of them. I’d also like a distinction between arable and pasture, but that would leave a residue of farmland sites not falling into either category.

We have a scheme available for iSpot use or heirarchical habitats, configured to the overarching IUCN habitats. Unfortunately, the programmers are not interested in implementing this. It would be a cool way of allowing southern African’s to include the national vegetation types for those capable (most have trouble at the biome level of merely Afromontane Forest, Eastern Subtropical Forest, Albany Thicket, Grassveld, Moist Savanna, Dry Savanna, Succulent Karoo, Nama Karoo, Fynbos, Renosterveld and Strandveld), but would also allow specialists to go finer to the community level.
Then your farmland could be arable, pasture (and even “other” for future unpacking), and Urban Parkland could be divided into several other types.

There is a southern African forum item addressing this, and the mismatches between the UK habitats and southern African habitats and how to resolve them, but it vanished in this rebuild.

I’m fairly sure that this topic was discussed on the forums a few years back.

As habitats are implemented as tags, there’s no reason why the users can’t be presented with an expandable menu, nor why different menus can’t be used for the different communities. It’s a bit more complicated for filtering reports, but it boils down to an a list of ORs on a database query, which is not rocket science.

1 Like

Different menus are available for different communities. The problem is that they are not properly integrated. A hierarchical system is thus better (i.e. the habitats need a taxonomy ((the same is true for interactions))).

The big problem comes for users who want to cross communities. Integrating the taxonomy and habitats is an issue that iSpot has not dealt with at all.
The way I see it, there should only be one dictionary on iSpot, with communities viewing their relevant taxa. Similarly, there should be one habitat dictionary, with communities viewing their relevant habitats. Different dictionaries and different habitats are not the way to go, especially if we are going to encourage new communities to join.

The advantage of a heirarchical system is that Hong Kong could have 20 different “Urban Parkland” and southern Africa just one type, and yet a filter could access this with just a single tag name and without the need for ORs and OR NOTs.

Agree, at least a few more would not hurt, one for ‘urban area’ would seem desirable to me.

We’d probably want to stop short of NVC though (http://jncc.defra.gov.uk/page-4259)

Stop short of NVC? No why? I disagree?

I think that requiring people to do NVC will be totally inappropriate, if possible. But having the capacity to allow those who do want to use it is certainly to be encouraged. However, it needs to integrate hierarchically with other classifications, which may take some effort

(Having just integrated our southern African vegetation types and Marine ecosystem types with the IUCN for our national Red List and Conservation Assessment databases, it is not that impossible).