Epstein's suicide should make us ask: Why don't we have cameras in every prison-cell? An analysis

A high number of prisoners are murdered, assaulted/raped, or commit suicide in prison and jails. This problem could be substantially addressed with basic technology to provide surveillance on ALL prisoners: As a ballpark estimate, at the higher end, it could cost $400M, ~$100M for up-front costs and $~300M for operating (that would likely decline over time). And that's assuming that today there are no cameras (which there are). There are privacy concerns to deal with; however, this article is just on the economics of such a camera in each cell policy.

I think it should be a top priority to do so.

We should care about prisoners:

The problem of prisoner-conditions is swept under the rug, because society doesn't care much about these people - they have been stripped of their full rights as citizens. But they are human beings, some of them are innocent of their crimes, and none of them should be judged solely on the worst thing they have done in their life.

In Epstein's case no cameras caught him in the act because cameras only "scan common areas and outside the prison, not inside the cells."

Be it resolved: There should be cameras in every prison-cell in the country, by law.

These cameras should be monitored remotely, in-person and by AI to monitor for anomalous behavior and facial ID to track by location. I am a big privacy person, but not in prison.

An analysis:

(I intend to update these numbers if people provide revisions)

I. Up-front Costs Of $100M

A) There are 2.3M prisoners in the United States, and fewer than that number of prison-cells. Let's assume our goal is 1 camera for every 1 prisoner - this is probably overkill - but it's the upper-limit of our needs

B ) The ZOSI 720P CCTV Camera costs $15 on Amazon for the camera - that includes motion detection, cloud storage and night vision. Let's say you negotiate that to $14, a very small discount.

C) Installing these cameras will require time and cost, wiring up the locations for electricity, using a secure network etc. Let's assume that costs $28 per camera install - double the cost of the camera - to test the upper-limit of cost.

One-time $100M cost to buy equipment and for installation

II. Operating Costs

A) Replacement cameras, let's assume these things break all the time given the environment and you need to replace it once a year. But since the wiring has already been installed, we will say the cost is only $30 total, $15 for equipment and $15 for install = $69M/year

B ) Cloud storage, while the camera comes with cloud storage, ours would need to be secure and potentially store for several years. With current "AWS Infrequent" costs (not the cheapest option but in-between), we are at $.0125 per GB.

If we assume 100GB for HD storage of video, that's ~$15 a year per prisoner. (This is at the higher end of most estimates) = $30M/year (and declining every year)

C) AI - you will want AI tools to be analyzing the video to spot anonymous behavior, hurting someone, cellphones, drugs, not breathing etc. This you outsource to a vendor and pay for use. Given current technology, this area currently has strong offerings in Intel community but would need to be modified for this arena. Assume that the AI offerings are limited in 1-2 years, and advanced in 3-5 years. Cost?

Since it's software it scales, you are probably looking at ~$100M for minor modifications of existing software (mainly from intel), but you can put this up for competitive bidding (and declining prices overtime).

D) You may have costs for watching live-streams, but for most part you are reducing other costs. Less frequent walks of the facility for example. With more surveillance you will have less incidents, and all sorts of savings.

Nonetheless, it's reasonable to assume that you will have to hire more people to watch and forward the video. I'd allocate $100M to be sure.

III. Other benefits

There would be many other positive benefits of this technology, cooperation with police/prosecutors, fewer re-offenders, higher safety for visitors, fewer phones in prison, reduced corruption of officers, prosecution for crimes in prison etc.

Further, advancing this technology could have military and border implications as well.

IV. Estimate Cost:

Again this is a ballpark estimate, but the point is to show that this is a reasonable proposition:

-Up-front costs: ~$100M

-Operating costs: ~$300M/year (and declining)

Let's say it's possible the numbers are double this - would $800M be too high a price to pay?

I'm interested in other recommendations and thoughts on cost, and will revise with feedback.