Geotagging on Flickr: flaky

Geotagging on Flickr

Flickr launched its geo­t­ag­ging feau­ture a few days ago. Today I came across a few rav­ing posts on Tech­n­Crunch, so I decid­ed to give it a go.

I’ve been geo­t­ag­ging my pho­tos using Plazes for a while now (has it been more than a year already? This pho­to seems to prove as much.) I enjoyed doing that but it was always a bit involved. Also, geo­t­ag­ging becomes real­ly use­ful and fun once lots of peo­ple start doing it. That wasn’t real­ly hap­pen­ing yet so I’m excit­ed about Flickr inte­grat­ing it.

My first impres­sion of their map-dri­ven inter­face was pos­i­tive. It’s tucked away in the orga­nize sec­tion though; I won­der whether they’ll include some bits in the indi­vid­ual pho­to pages soon. For instance: a lit­tle map show­ing the loca­tion where the shot was tak­en and an easy way to add geo­t­ags (maybe even allow oth­ers to do it for me?) I’d like this most­ly because now the map isn’t real­ly social (in the sense that it shows an aggre­ga­tion of geo­t­agged shots, just my own.) Update: I found the social fla­vored map here; a bit under­whelm­ing, but fun.

How­ev­er: although Flickr proud­ly sports “gam­ma” at the top of its logo, the tech­nol­o­gy still lags behind. It’s beta qual­i­ty at best. New­ly tagged pho­tos don’t appear on the map after a reload; per­haps Flickr doesn’t like me chang­ing the tags out­side of the map inter­face? Update: edit­ting geo­pri­va­cy set­tings on batch­es gives back strange results too, these pho­tos should show up on the map some­where near Baarn, but they don’t. Weird…

Also, I think not being able to “snap” a batch of pho­tos to a city I found through the search inter­face is a usabil­i­ty issue. Adding pho­tos to loca­tions I haven’t iden­ti­fied in Plazes (and thus don’t show up as hotspots on the map yet) becomes arbi­trar­i­ly. Call me a meta­da­ta nut, but I real­ly want to add my pho­tos of Jurjen’s pret­ty street Zwarte­hand­spoort in Lei­den to the exact street, not drop them some­where in the vicin­i­ty of the city Lei­den.

Con­clu­sion: a promis­ing addi­tion to every­ones favourite social pho­to shar­ing site, poised to make geo­t­ag­ging ready for the big time, but not exact­ly there yet due to some tech­ni­cal and design issues.

Anoth­er update: after rum­mag­ing through the help forums, I learnt that indeed, Flickr doesn’t automag­i­cal­ly pick up on new­ly geo­t­agged pho­tos from oth­er ser­vices (such as Plazes.) You need to re-import them (as described in this post). This sucks big time, Flickr seems to think that only pho­tos that have been tagged inside the sys­tem mat­ter. Think again! (Of course all this is prob­a­bly sim­ply due to tech­ni­cal lim­i­ta­tions, which is no excuse, but still…)

Rough notes for Jesse James Garrett — Keynote

Adap­tive Path intro

  • Big & small clients
  • Ele­ments of UX: “under­stand­ing tool”
  • the AJAX guy”

Prob­lem

  • We know lit­tle about peo­ple, hard to make good guess­es
  • IA is about find­ing ways to make bet­ter guess­es, but they’re still guess­es

Tech­niques:

  • Card sort: prim­i­tive, low tech
  • New approach: just give up, cre­ate a sys­tem for users to cre­ate their own archi­tec­ture (tag­ging)
  • Prob­lems (no such thing as mag­ic): insid­er lan­guage, con­trolled vocab­u­lary non exis­tent, most pop­u­lar is not nec­es­sar­i­ly the best, tag spam, tag­bomb­ing

How to improve tag­ging? First step towards user gen­er­at­ed IA From explic­it IA woes to implic­it user gen­er­at­ed IA

Exam­ple: Ama­zon

  • Algo­rith­mic archi­tec­ture
  • Indi­vid­ual and aggre­gate data com­bine to cre­ate gen­er­at­ed IA

Next step for algo­rith­mic archi­tec­ture

  • use them in the right place
  • make them trans­par­ent to the user

Bet­ter data

  • Two fla­vors: about con­tent (meta­da­ta) and about users (now: user research, in thee future)
  • Usabil­i­ty test­ing is like blind man’s cane
  • Bet­ter canes aren’t the answer, make the blind see
  • Instru­ment­ed inter­faces: hav­ing a site be a con­tin­u­ing exper­i­ment and feed back data to design­ers
  • Exam­ple of Ama­zon URL: domain, CMS junk, prod. ID, inter­face tag, ses­sion ID
  • Inter­face tag tells you where users were click­ing on a page
  • Search results: use query ID, tells you about search terms used

Exam­ple of base­ball sta­tis­tics

  • Start sim­ple, then go to basic math, then to com­plex cal­cu­la­tions

Get­ting data isn’t enough

Ethics

  • Sep­a­rate behav­ioral data from Influ­ence cor­po­rate pol­i­cy

Ear­ly days

  • Real poten­tial is still untapped, we need bet­ter ana­lyt­ics tools

Ques­tions Q Col­lab­o­ra­tive fil­ter­ing only tells you what choic­es were made, not what all pos­si­ble choic­es were? A How do we pre­serve serendip­i­ty, don’t get locked in feed­back loops? Peo­ple have been work­ing to rein­tro­duce serendip­i­ty.

Q How do we pri­or­i­tize com­mon knowl­edge about com­mu­ni­cat­ing vis­it­ed links? A Is it nec­es­sary to com­mu­ni­cate it? He thinks it’s an open ques­tion.

Q Is Ama­zon now in the busi­ness of push­ing this tech? A They’re cer­tain­ly mov­ing in the direc­tion…

http://reboot.dk/wiki/BeyondTag­ging:User-Gen­er­at­edInfor­ma­tionArchi­tec­ture

Next generation tag clouds

Joe Laman­tia recent­ly pub­lished a two-part essay on tag clouds. He’s man­aged to cre­ate a nice primer on tag­ging from an IA’s point of view. His tone of voice is a bit aca­d­e­m­ic, which may turn you off, but his pre­dic­tions of future direc­tions for tag­ging and tag clouds are spot on quite inter­est­ing.

A while ago I tried to come to grips with the tag­ging phe­nom­e­non in a series of posts on open.info.nl (in Dutch). The last one was on facetted tag­ging. Laman­tia briefly men­tions this as a new direc­tion, but doesn’t real­ly describe how he imag­ines this would work.

I’m in the plan­ning phase of a facetted tag­ging demo, and still not sure how to approach it: should con­tent edi­tors spec­i­fy the facets with­in which con­tent users can tag, or should the facets be tags as well. In oth­er words – how far should we go in relin­quish­ing con­trol over meta­da­ta?

Ondergrond.org — HKU-studenten aan de folksonomy

Via com­men­taar op een recent artikel op open.info.nl kwam ik op de site Onder­grond – een folk­son­o­my voor / van street art. De site is een EMMA-afs­tudeer­pro­ject van een aan­tal HKU-stu­den­ten. De site daagt bezoek­ers met behulp van stellin­gen en vra­gen uit om bij foto’s van graf­fi­ti en stick­ers tags achter te lat­en. Een inter­es­sante manier om het dilem­ma “waarom zou een bezoek­er taggen” te tack­e­len – het principe doet me in die zin denken aan Hot or Not. Het plezi­er zit hem in foto na foto hersen­loos te voorzien van meta­da­ta. Het risi­co is natu­urlijk dat hier­mee het ontstaan van “metacrap” alleen maar in de hand wordt gew­erkt! Aan de andere kant zijn de vra­gen soms wel wat moeil­ijk, dan moet je goed nadenken, en is het effect van de laag­drem­pe­ligheid weg.

Ik weet niet of Maarten en Sjors Inter­ac­tion Design hebben ges­tudeerd, maar op dat vlak ver­di­ent de site wel nog wat aan­dacht. Het is flink zoeken geblazen in het onder­grondse, de nav­i­gatie is eigen­lijk bij­na niet aan­wezig. Miss­chien dat dit niet de focus heeft in hun project, maar het zou toch mooi zijn als het de tag­ger makke­lijk wordt gemaakt zijn weg te vin­den naar inter­es­sante con­tent!

Edgeio launches

Devel­op­ment of an open plat­form for list­ings con­tin­ues with yesterday’s launch of Edgeio. This ser­vice picks up on blog post that are marked as a list­ing using a Micro­for­mat-like tag and col­lects them on a web­site for users to search through. They have some nifty inter­faces in place to allow you to zoom in on list­ings that are close to you. One thing they haven’t addressed yet (but are aware of) is the prob­lem of fake list­ings and spam.

Struc­tured Blog­ging is work­ing in the same field, but from a dif­fer­ent angle (focussing on the blog­gers, not the site that col­lects list­ings). I wrote about them ear­li­er over here (in Dutch). The biggest chal­lenge for the Edgeio crew is prob­a­bly gain­ing crit­i­cal mass to be able to com­pete with the likes of eBay.

Every­one was blog­ging about this yes­ter­day, here are some of the posts:

  1. TechCrunch » Edgeio Launch­es
  2. Read/WriteWeb: edgeio launch­es with same goals as Struc­tured Blog­ging
  3. WeBreak­Stuff » Edgeio launch­es
  4. GigaOM : » Edgeio Launch­es, Final­ly
  5. Rus­sell Beat­tie Note­book — Play­ing with EdgeIO

Screens in train – again

Nou, ein­delijk die WiFi van de NS kun­nen testen. En hij werkt best aardig (op de spo­radis­che time out na). Gelijk een dilem­ma op Plazes. Hoe geo­t­ag je een bewe­gend object? Beet­je jam­mer dat ze het alleen niet voor elka­ar kri­j­gen de juiste reis­in­for­matie op de scher­men te kri­j­gen. Op weg naar Den Bosch is vol­gens de NS de vol­gende halte Ams­ter­dam Cen­traal, tot grote ver­war­ring van menig reiziger…

NStv testtrein