Recent diary entries
I removed this diary entry.
I was planning to complete a few edits over this weekend. However, before I could start mapping I have to monitor recent edits in my region to see what people are doing. OSM has no kind of safeguards for accidental deletions and other types of errors. It all falls down to individual users monitoring their areas and to close the gate after the horse escaped.
In my previous diary entry I reported on edits by new users in Brazil which are damaging the map much faster than we can fix. This weekend seems to be no exception.
I am the only user in a metropolitan region of over 3 million people who watches this area. Apart from me there is only one other active mapper in this area, but he does not use JOSM and would not know how to revert edits or to recover deleted object. Unfortunately, I cannot watch user changes every day, so it may take a while until I find out damages. The longer it takes to find, the harder it becomes to undo those damages.
Already the first new user in the area I watch on his/her very first edit deleted some administrative boundarys. As usual, no reasons were given (entry is "no comment"). I am not comfortable at all using JOSM's reversion plugin, but it seems I was successful in reverting this time.
The second new user also deleted several ways, although it seems it was something she created a few days earlier. This will need further and laborious investigation. Again, no reasons were given. I've contacted the user, lets hope she replies (they usually don't).
The third user is not exactly a new one, he started mapping about two months ago. None of his 11 edits have comments. But he too deleted boundaries in his first edits which were kindly fixed by our colleague naoliv. I had contacted this user about a month ago, but he still writes no comments in his edits. I see that nothing was deleted this time, but I discovered that there are still deleted elements of earlier edits. Let's hope I manage to fix them, after 2 months there may be tons of conflicts already.
I obviously can not check in detail every single edit by users in my area. I verify only deletions because these are the hardest to recover once they are gone. Any other mistakes are likely to stay for a very long time.
On a national level, there appeared 22 new users since Tuesday. As for the local 3 new users, I will send each one a welcoming message in Portuguese. Sending 22 messages over OSM's messaging system is quite a tedious tasks. But is has to be done. Our only hope to keep damages at a minimum is to contact theses users as soon as possible.
Note: this diary entry is an effort of the Brazilian OSM community to convince our colleagues from developed countries that the complete lack of preventive actions on new users is overwhelming the more experienced mappers in this part of the world.
It is a real shame that OSM provides no level of protection to any element the map. The map is growing not only in size but also in complexity. It is very easy to damage things. For example, deleting a single node may render a complex relation, like a turn restriction, completely useless.
In this example a newly registered user in few minutes and on his very first edit damaged a boundary relation with over 5 years of history and on which 5 experienced users work very hard. This shouldn't be so easy in the first place.
There is no doubt that this user never knew his was breaking something. Very likely he got no warning whatsoever from iD, or if he got one, he did not understood what was going on.
It is also very likely that this user will make no further contribution to the map, not even a bad one. The vast majorities of user add one or two edits and never return.
The damage which this user inflicted is a small one, but it is no isolated incident. It is happening all the time and is getting worse.
Hundreds of new users are arriving on a daily basis. Each one is granted immediate access to the base with no restriction of any kind . This may have make sense in the early days of OSM, but not any longer. It is not even helpful for the new users. There is nothing more discouraging for a new user, who is eager to help, to discover that he or she broke something accidentally. And quite frankly iD is not helping a single bit in this respect.
Since we cannot block data from being damaged, the only sort-of-workaround is to use third party services, like whodidt, to monitor changes to the map continuously. But these services only alert us when the damage has already been done, like in the example above. Besides, even then it is sometimes very hard to spot the damage and correcting it is sometimes more time consuming then the original mapping. The fact that there are no built-in systems to alert us of critical changes also speaks for itself.
In the particular case of Brazil, where there are only a few active mappers (something between 20 to 30) this is proving a real burden, especially considering the size of the country. To put this in perspective, it would be like having just 2 mappers for Germany. Yet the weekly influx of new users in Brazil is around 30 to 50. Almost each day something is being broken, sometimes very badly like in this case were a new mapper deleted elements affecting all of South America.
My feeling is that we can no longer go on like this, and I am quite surprised that the companies who rely on OSM for their businesses are not crying out already.
OSM is often portreayed as the wikipedia of maps. Yet wikipedia, which once was completely free to edit, had to undergo some difficult changes.
I think that it is high time that OSM shifts it priority from gathering tons of new users to preserving the integrity of the map. Not only will we have a more reliable map, but we also will free the experienced mappers from the burden of constantly watching and correcting things that were not broken before.
The longer we wait, the harder it will get.
Estou revisando algumas estradas na Bahia, principalmente na região de Ilhéus.
Há um acervo de dados muito ricos do DER-BA, que de acordo com o próprio site "Todo o conteúdo deste portal pode ser utilizado livremente, desde que a fonte seja citada."
Portanto existe aqui a possibilidade de colocar as denominações corretas das rodovias da Bahia. Muitas delas estão com códigos errados, a maioria não tem código de referência algum.
Informações que eu tenho colocado:
surface=paved ou surface=unpaved pelas próprias informações do DER-BA (source:surface=DER-BA) ou pelo que pode ser visualizado pelas imagens de alta resolução do Bing.
source=IBGE;DER-BA;High Res Bing quando se trata de uma combinação de informações e a via pode ser traçada pelas imagens de alta resolução
Decidi mapear a Serra da Canastra da maneira mais completa possível: todas as vias, trilhas, rios, córregos, serras, matas, fazendas e cidadezinhas. ]
Porque? Primeiro que é uma das regiões de maior valor turístico em Minas Gerais e não tem mapa nenhum que presta. Diferente de cidades grandes e capitais, sempre há alguma alternativa em termos de mapas. Nessas regiões não existe nada, ou seja, há uma carência muito grande de mapas.
O segundo motivo é porque mapas mais completos impressionam e motivam. Hoje em dia, fora das regiões urbanas o que nós temos no OSM-Brasil são grandes vazios. Eu quero criar um mapa bonito da Serra da Canastra e seus arredores que vire referência para todos na região. Como você vê, vou passar muito tempo fazendo isto :)
A ferramenta principal que estou usando é o JOSM e o plugin Piclayer. O que faço é pegar os mapas pdf do IBGE e superponho com as imagens de satélite do Bing, veja o procedimento (aqui)[https://github.com/nighto/calibracao-mapas-ibge]. Assim eu acho todos os rios e córregos (em geral encobertos por vegetação) bem como o nome deles.
Estive recentemente em Sítio do Conde e verifique a completa ausência de qualquer mapeamento da área e a ausência de dados do Bing. Percorri algumas áreas com o meu GPS e fiz um mapeamento incial, concentrando principalmente em Sítio do Conde.
Também ajustei (mas não muito) a localização do trevo da BA-099 e BA-233.
Finalmente removi um track que havia entre a BA-099 passando por Conde e indo na direção de Sítio do Conde que não existe. A fonte informada para este track é do IBGE, mas não acredito que tenha existido algum dia pois atravessaria o rio e passaria por uma área completamente alagada.
Adicionadas a Rua Dona Adelaide e a Rua Retiro no Caiçara, BH que estavam faltando.
Corrigido o traçado da BR381 entre Nova Venécia e Rio Preto a partir de dados de GPS que coletamos no dia 01/02/2012 e 14/02/2012. Esta área ainda não está coberta pelo Bing e traçado anterior era completamente irreal.
Ajudou muito o uso do mapa do DER-ES para saber a referência correta das estradas nesta parte do espírito santo.
Adicionei as duas estradas que levam a Cumurxatiba: a estrada litorânea que inicia em Prado, muito bonita mas que deve ser evitada em dias de chuva a partir dos meus dados de GPS. Aproveitei para colocar o trecho que falta da BA001 a partir das imagens do Bing.