In January 2017, a Swedish hacker named Linus Trulsson created an internet robot to manipulate Numbeo's crime survey data, resulting in Lund being falsely portrayed as the top city in the world in Numbeo's Current Crime Index. The incident was reported by TheLocal.se , a Swedish media outlet. The hacker utilized an exploit that allowed unlimited users from the same IP address on Numbeo. Once this manipulation was brought to our attention, we promptly addressed the issue by closing the exploit within 24 hours and marking all inputs from the user in Sweden as spam.
In 2022, Numbeo's data for Bradford was manipulated by an unidentified individual or group who took advantage of VPN networks, public proxy servers, and changes in dynamic IP address assignments of a router based in Greece. This incident gained widespread attention in the UK press, with the police dismissing the claim. Upon receiving the report, we initiated an immediate investigation and subsequently flagged specific IP addresses as spammers. To minimize the likelihood of future manipulations on our database we made improvements in our code to automatically detect attempts of data manipulation by the usage of VPN networks and ranges of IP addresses. Data for Bradford were manipulated by usage of VPN network and usage of dozens of IP addresses that all belong to dynamically assigned IP address of the same ISP providers.
We want to ensure a community that we are determined to fight data manipulation attempts by using multiple IP addresses or otherwise. As explained in our Methodology document, we reserve the right to flag certain IP addresses as spammers. Once those IP addresses are flagged, we don’t use their inputs in our calculations, while we are using their inputs to detect future manipulation attempts. We are using this cautiously, as we don’t want to discredit legitimate inputs that help us build this database.
We are determined to allow anonymous inputs without registration, as enforcing registration could lower the number of inputs our database gets and reduce accuracy in certain ways. As our inputs are numerical, we believe algorithms can be constructed to fight spammers.
We want to add that there are multiple positive or neutral mentions on our data and methodology, and we want to illustrate it with a few examples (more examples could be found in Google Scholar or Google News).
Examples:
@incollection{jayadev20183,
title={3. The Middle Muddle: Conceptualizing and Measuring the Global Middle Class},
author={Jayadev, Arjun and Lahoti, Rahul and Reddy, Sanjay},
booktitle={Toward a Just Society},
pages={63--92},
year={2018},
publisher={Columbia University Press}
}
@article{helble2021affordable,
title={How (Un) affordable is housing in developing Asia?},
author={Helble, Matthias and Ok Lee, Kwan and Gia Arbo, Ma Adelle},
journal={International Journal of Urban Sciences},
volume={25},
number={sup1},
pages={80--110},
year={2021},
publisher={Taylor \& Francis}
}
If you have additional question about these matters, please contact us.
Last Update: May 19th, 2023.