Othram - General Discussion - #3

[Caring1]

Introducing Solvepedia (v0.1)

We’ve started testing something new: Solvepedia.org. It’s very early (v0.1), but the idea is to build a fully autonomous system that can learn, scale, and help expand what we know about missing and unidentified persons.

Wikipedia is an incredible resource, but it relies on manual contributions and strict notability rules, so many cases never get documented or don’t stay published. We’re exploring a different approach.

Long-term, we hope this system can:
• Collect and organize information across cases
• Continuously improve and expand coverage
• Potentially help surface connections between missing and unidentified persons

Would love your feedback. You can contribute directly or highlight text on any page to comment.

Take a look and let us know what you think.

https://solvepedia.org/

A search tool that could give one the option to automatically include all words that mean the same. A synomatic function of sort. Just made that up but if it could pick up all synonyms when you are searching a specific description.

For example, while searching acne, the search would also automatically include pockmarks, blemishes, etc.

Or for clothing, swimsuit would automatically include bathing suit, swim trunks, swim shorts, etc.

Necklace would include, gold chain, yellow metal chain, silver chain, white metal chain, etc.

Tennis shoes would also automatically include sneakers. I could go on but these are just some examples that could condense many searches into one or at least have the option to do so.

 

[othram]

A search tool that could give one the option to automatically include all words that mean the same. A synomatic function of sort. Just made that up but if it could pick up all synonyms when you are searching a specific description.

For example, while searching acne, the search would also automatically include pockmarks, blemishes, etc.

Or for clothing, swimsuit would automatically include bathing suit, swim trunks, swim shorts, etc.

Necklace would include, gold chain, yellow metal chain, silver chain, white metal chain, etc.

Tennis shoes would also automatically include sneakers. I could go on but these are just some examples that could condense many searches into one or at least have the option to do so.

This is the future!

 

[Caring1]

This is the future!

So looking forward to this. Cannot wait to explore.

 

[iulia]

Would love your feedback. You can contribute directly or highlight text on any page to comment.

Take a look and let us know what you think.

https://solvepedia.org/

RSBM

Hello! Just wanted to throw in a suggestion when it comes to the NamUs links on Solvepedia: you might want to use the namus.gov links to cases instead of namus.nij.ojp.gov ones. They are different platforms and the former is the “original”, accurate one. The latter gets hourly feeds from namus.gov, but it is prone to errors/glitches. I contact NamUs once or twice a year because it stops working properly or stops working altogether, sometimes for long periods of time. I don’t know the tech jargon, so I don’t know if I am able to properly explain the issue. I’ll paste part of a message I once got from NamUs, in case it helps understand the difference between the two:
“The public-facing pages of NamUs, located at https://namus.nij.ojp.gov, are supported by a Department of Justice Data Management, Reporting and Analytics system which receives hourly feeds from the NamUs transactional database. As a feeder system, the NamUs transactional database will always be more current than the system that supports public pages […].”

To exemplify, the first page I opened on Solvepedia and checked the sources links 3 times to a namus.nij.ojp.gov page that doesn’t exist. It should exist, as the case is still open on namus.gov, but in this case, for unknown reasons, it doesn’t work.

Waller County Jane Doe (1991)

Case: UP100484

solvepedia.org

https://namus.nij.ojp.gov/case/UP100484

The National Missing and Unidentified Persons System (NamUs)

By bringing people, information, forensic science and technology together, NamUs helps resolve cases.

www.namus.gov

 

[SimsGuy67]

No humans. Fully autonomous.

Sorry, but IMO this isn't really something to brag about. You can talk about how AI is the future all you want, but when it comes to very detailed cases and such, is it really smart to trust AI when it can make even the simplest mistakes?

Like, imagine if someone goes onto the site looking for a possible match to one of their relatives, and finds a possibility, but they disregard it due to incorrect information the AI has put on the entry.

Not to mention it discredits the real work dozens of people truly committed to helping solve these cases have done. Especially in more sensitive cases, you have to be so careful not to come off as insensitive or spread misinformation, and that's even for humans. I doubt AI, in its current form, can do this.

With all due respect, I can't support this at all, and I think it could do more harm than good.

 

[othram]

Sorry, but IMO this isn't really something to brag about. You can talk about how AI is the future all you want, but when it comes to very detailed cases and such, is it really smart to trust AI when it can make even the simplest mistakes?

Like, imagine if someone goes onto the site looking for a possible match to one of their relatives, and finds a possibility, but they disregard it due to incorrect information the AI has put on the entry.

Not to mention it discredits the real work dozens of people truly committed to helping solve these cases have done. Especially in more sensitive cases, you have to be so careful not to come off as insensitive or spread misinformation, and that's even for humans. I doubt AI, in its current form, can do this.

With all due respect, I can't support this at all, and I think it could do more harm than good.

Really appreciate you taking the time to share this.

The goal here isn’t to replace human expertise or the incredibly important work that people have been doing for years. If anything, it’s the opposite. The idea is to support that work by surfacing connections, patterns, or leads that might otherwise be missed simply because of the magnitude or scale of the problem (too many unidentified and missing folks).

The reality is that we can’t realistically scale human effort to continuously compare every missing person case against every unidentified person case. There are just too many records, and more are added all the time. If we want to improve how often and how quickly these cases are resolved, we do need technology to help expand what’s possible.

So this is a research project to explore where technology can help. This is not unlike Google or Apple Maps. Both useful tools that were not so useful when they first launched

 

[othram]

RSBM

Hello! Just wanted to throw in a suggestion when it comes to the NamUs links on Solvepedia: you might want to use the namus.gov links to cases instead of namus.nij.ojp.gov ones. They are different platforms and the former is the “original”, accurate one. The latter gets hourly feeds from namus.gov, but it is prone to errors/glitches. I contact NamUs once or twice a year because it stops working properly or stops working altogether, sometimes for long periods of time. I don’t know the tech jargon, so I don’t know if I am able to properly explain the issue. I’ll paste part of a message I once got from NamUs, in case it helps understand the difference between the two:
“The public-facing pages of NamUs, located at https://namus.nij.ojp.gov, are supported by a Department of Justice Data Management, Reporting and Analytics system which receives hourly feeds from the NamUs transactional database. As a feeder system, the NamUs transactional database will always be more current than the system that supports public pages […].”

To exemplify, the first page I opened on Solvepedia and checked the sources links 3 times to a namus.nij.ojp.gov page that doesn’t exist. It should exist, as the case is still open on namus.gov, but in this case, for unknown reasons, it doesn’t work.

Waller County Jane Doe (1991)

Case: UP100484

solvepedia.org

https://namus.nij.ojp.gov/case/UP100484

The National Missing and Unidentified Persons System (NamUs)

By bringing people, information, forensic science and technology together, NamUs helps resolve cases.

www.namus.gov

Good catch. Thank you for surfacing this!

 

[othram]

Nearly 50 years after his remains were discovered, the Blount County Sheriff’s Office, the University of Tennessee Forensic Anthropology Center, and the Tennessee Bureau of Investigation never gave up, working together and using Othram technology to identify a 1981 John Doe as William Thomas Green.

After 44 Years, Blount County John Doe (1981) is Identified

 

[dotr]

Nearly 50 years after his remains were discovered, the Blount County Sheriff’s Office, the University of Tennessee Forensic Anthropology Center, and the Tennessee Bureau of Investigation never gave up, working together and using Othram technology to identify a 1981 John Doe as William Thomas Green.

After 44 Years, Blount County John Doe (1981) is Identified

D

TN Thread 'TN - East Miller's Cove area of Blount County, Knoxville John Doe, #UP1591, 4 April 1981, DNA - William Thomas Green (born March 29, 1942), 2026.'

Today at 12:13 PM

William Thomas Green

After 44 Years, Blount County John Doe (1981) is Identified

Blount County Sheriff's Office, University of Tennessee Forensic Anthropology Center, and Tennessee Bureau of Investigation Team with Othram to Identify a 1981 John Doe

dnasolves.com

''On April 4, 1981, a local hunter discovered the body of a man in the East Miller's Cove area of Blount County, Tennessee. Blount County is just south of Knoxville, Tennessee. Investigators responded to the scene and began collecting evidence and investigating. A forensic anthropology examination determined that the remains were those of a white man who was likely between the ages of 18 to 35 years old when he died, likely sometime between 1977 and 1980.

Investigators with...

• dotr

• 

 

[SimsGuy67]

Really appreciate you taking the time to share this.

The goal here isn’t to replace human expertise or the incredibly important work that people have been doing for years. If anything, it’s the opposite. The idea is to support that work by surfacing connections, patterns, or leads that might otherwise be missed simply because of the magnitude or scale of the problem (too many unidentified and missing folks).

The reality is that we can’t realistically scale human effort to continuously compare every missing person case against every unidentified person case. There are just too many records, and more are added all the time. If we want to improve how often and how quickly these cases are resolved, we do need technology to help expand what’s possible.

So this is a research project to explore where technology can help. This is not unlike Google or Apple Maps. Both useful tools that were not so useful when they first launched

Okay, but the thing is, this statement didn't really address my main concern, and that is that the AI may fabricate information, which, if anything, would bury real connections. When you apply the rate of how often AI gets stuff wrong, it could impede the investigation of hundreds of cases. Maybe thousands.

 

[othram]

Okay, but the thing is, this statement didn't really address my main concern, and that is that the AI may fabricate information, which, if anything, would bury real connections. When you apply the rate of how often AI gets stuff wrong, it could impede the investigation of hundreds of cases. Maybe thousands.

We are not using AI to generate facts. The goal is to work from existing records and help surface possible connections across them, not invent new information.

Also, today's current system is not perfectly accurate. In reality, we already have thousands of missing and unidentified cases that are never compared against each other simply because it is not possible to do that manually at scale. That means real connections are already being missed, not because of AI errors, but because the work cannot physically be done.

Even when comparisons are made, they are not error proof. Anthropological and other forensic analyses have limitations as well, and connections are often missed. Look at the Little Miss Nobody case or Beth Doe or countless cases where ambiguous or incorrect facts or guess derailed the case.

And more broadly, none of the current information channels, whether forums, social media, or community discussions, are immune to inaccuracies either. That is true across the board.

What we are doing is asking a valid research question: can technology help us build something that is structured, traceable back to sources, and designed to support review and contribution? Because the status quo, where many cases are never meaningfully compared at all, is not good enough. Building a data system that can traceback to sources and accept and implement feedback from the community is a powerful tool.

Doing nothing is not a better option.

 

[SimsGuy67]

We are not using AI to generate facts. The goal is to work from existing records and help surface possible connections across them, not invent new information.

Also, today's current system is not perfectly accurate. In reality, we already have thousands of missing and unidentified cases that are never compared against each other simply because it is not possible to do that manually at scale. That means real connections are already being missed, not because of AI errors, but because the work cannot physically be done.

Even when comparisons are made, they are not error proof. Anthropological and other forensic analyses have limitations as well, and connections are often missed. Look at the Little Miss Nobody case or Beth Doe or countless cases where ambiguous or incorrect facts or guess derailed the case.

And more broadly, none of the current information channels, whether forums, social media, or community discussions, are immune to inaccuracies either. That is true across the board.

What we are doing is asking a valid research question: can technology help us build something that is structured, traceable back to sources, and designed to support review and contribution? Because the status quo, where many cases are never meaningfully compared at all, is not good enough. Building a data system that can traceback to sources and accept and implement feedback from the community is a powerful tool.

Doing nothing is not a better option.

Yeah, you're not using AI to generate facts deliberately, but the AI can easily make up stuff on its own. As you said, "No humans. Fully autonomous." I'm not saying humans don't make errors when researching or investigating cases, but I'm a bit confused as to why you think bringing in something with the opportunity to make even more errors will help the situation.

But, I guess we'll just have to agree to disagree on this. No matter what I say, this is going to go through anyway, so why waste my breath?

 

[Grungster]

Yeah, you're not using AI to generate facts deliberately, but the AI can easily make up stuff on its own. As you said, "No humans. Fully autonomous." I'm not saying humans don't make errors when researching or investigating cases, but I'm a bit confused as to why you think bringing in something with the opportunity to make even more errors will help the situation.

But, I guess we'll just have to agree to disagree on this. No matter what I say, this is going to go through anyway, so why waste my breath?

Plus, AI is very bad for the environment. Water and electricity. Not too familiar with that problem but need to do more research on it. AI is also just terrifying...

 

[maamager]

the whole “ai can make stuff up” is just a different version of the same problem we already have, humans miss connections all the time because there’s just too much data to cross-reference manually. that’s the baseline issue.

the point of this isn’t to let ai generate facts or replace investigators, it’s to act as a pattern-finding layer across existing, source-backed info. if it surfaces a potential connection, cool, i imagine that humans still verify it. if it’s wrong, it gets tossed. nothing is blindly accepted?

so the real question isn’t “can it make errors,” it’s “can it help surface leads we’d otherwise never even look at?” because right now, a ton of those just get missed entirely.

the environmental point is the most concerning... but that applies to basically all modern tech (search, streaming, social, etc). at that point it’s more about whether the use case is worth it, and organizing cold case data feels like a higher-value use than most.

 

[othram]

the whole “ai can make stuff up” is just a different version of the same problem we already have, humans miss connections all the time because there’s just too much data to cross-reference manually. that’s the baseline issue.

the point of this isn’t to let ai generate facts or replace investigators, it’s to act as a pattern-finding layer across existing, source-backed info. if it surfaces a potential connection, cool, i imagine that humans still verify it. if it’s wrong, it gets tossed. nothing is blindly accepted?

so the real question isn’t “can it make errors,” it’s “can it help surface leads we’d otherwise never even look at?” because right now, a ton of those just get missed entirely.

the environmental point is the most concerning... but that applies to basically all modern tech (search, streaming, social, etc). at that point it’s more about whether the use case is worth it, and organizing cold case data feels like a higher-value use than most.

Yes, of course. The point is to organize information and surface leads for humans to act upon.

 

[othram]

Yeah, you're not using AI to generate facts deliberately, but the AI can easily make up stuff on its own. As you said, "No humans. Fully autonomous." I'm not saying humans don't make errors when researching or investigating cases, but I'm a bit confused as to why you think bringing in something with the opportunity to make even more errors will help the situation.

But, I guess we'll just have to agree to disagree on this. No matter what I say, this is going to go through anyway, so why waste my breath?

It is a better use of resources to work on solving that problem than speculating about worst case scenarios without testing whether a better system is possible.

 

[othram]

Nearly 49 years after his remains were discovered in Seminole County, Georgia, the Seminole County Sheriff’s Office and the Georgia Bureau of Investigation used Othram technology to identify a 1976 John Doe as Curtis Lee Jones.

After Decades, Seminole County John Doe (1976) is Identified

 

[Oakie]

Introducing Solvepedia (v0.1)

We’ve started testing something new: Solvepedia.org. It’s very early (v0.1), but the idea is to build a fully autonomous system that can learn, scale, and help expand what we know about missing and unidentified persons.

Wikipedia is an incredible resource, but it relies on manual contributions and strict notability rules, so many cases never get documented or don’t stay published. We’re exploring a different approach.

Long-term, we hope this system can:
• Collect and organize information across cases
• Continuously improve and expand coverage
• Potentially help surface connections between missing and unidentified persons

Would love your feedback. You can contribute directly or highlight text on any page to comment.

Take a look and let us know what you think.

https://solvepedia.org/

Great idea. You'll have to work hard to keep the chaff separate from the grain, but don't let that stop you. If the effort proves that it "has legs", you'll get a lot of contributions from people with local knowledge.....