4 ms·

Here's the thing about the entire argument thread. It not being new is not reason enough for it to continue. It already existing does not mean we should never

by pempem 5y ago

Here's the thing about the entire argument thread. It not being new is not reason enough for it to continue. It already existing does not mean we should never revisit this decision. The technology on the other side of the feed is new and it has changed tremendously.

jcrawfordor 5y ago
Has it, though? While much discussed, facial recognition on surveillance is very expensive in practice and so uncommon to a degree I think many here would find surprising. The article here discussed facial imagery captured by handheld cameras, not involving the surveillance system. That's why I think it's important to split these hairs: video surveillance recording of faces for post-incident use is very different from proactive use of facial recognition. The former has been accepted for decades as a practical necessity for investigation of crime. The latter is new. I think that framing the discussion around cameras at face level, which is not new, misses the point and starts from a premise that will be wholly rejected by the industry. We need to focus instead on the use of automated technology to process this data, which is new and not yet widely practiced. That is changing. Newer self checkout terminals correlate video from both cameras with purchase history, primarily as a short-term local aid to staff when figuring out discrepancies. But this can obviously be extended to a long-term retention business intelligence application. The time is now to put pressure against this kind of data collection. But saying "there should be no cameras pointed at the face" will not apply that pressure. It will be dismissed because the use of cameras for face capture is a universally accepted anti-crime practice, and arguing against it as a way to object to mass surveillance ignores the differences between short term retention of data for post incident use and mass collection.

by pempem 5y ago

jcrawfordor 5y ago

Has it, though? While much discussed, facial recognition on surveillance is very expensive in practice and so uncommon to a degree I think many here would find surprising. The article here discussed facial imagery captured by handheld cameras, not involving the surveillance system. That's why I think it's important to split these hairs: video surveillance recording of faces for post-incident use is very different from proactive use of facial recognition. The former has been accepted for decades as a practical necessity for investigation of crime. The latter is new. I think that framing the discussion around cameras at face level, which is not new, misses the point and starts from a premise that will be wholly rejected by the industry. We need to focus instead on the use of automated technology to process this data, which is new and not yet widely practiced. That is changing. Newer self checkout terminals correlate video from both cameras with purchase history, primarily as a short-term local aid to staff when figuring out discrepancies. But this can obviously be extended to a long-term retention business intelligence application. The time is now to put pressure against this kind of data collection. But saying "there should be no cameras pointed at the face" will not apply that pressure. It will be dismissed because the use of cameras for face capture is a universally accepted anti-crime practice, and arguing against it as a way to object to mass surveillance ignores the differences between short term retention of data for post incident use and mass collection.