Deeming - A closer look

[The Rookie Detective]

Hi Michael,

I wasn't very clear in my intentions, but Lewis C has put it more clearly than I did. I used Long as an example because if we were to apply the "suspect" scoring system, then in terms of location score, she gets +2 (all that is required to get +2 is "be in London", and we know she was). Basically, almost everyone gets +2, except those where there might be some question as to whether or not the were in London at the time, in which case we're currently giving them +1 or 0. Basically, we score those whose location is questionable with a lower value than the default +2.

In my view, that makes the +2 meaningless, and the "information" of the score system is really contained within the lower scores we give those who are outside of London, where we give +1 if they could get there without too much difficulty, and 0 if getting to London would be sufficiently difficult. In my opinion, Sickert gets a 0 because he appears to have been in France, and while it is physically possible for him to board a boat and sail over, then return, etc, that was not as easy as it is today. Others, like Gull and Druitt, are within a fairly easy journey (Gull was renting a summer cottage or something outside of London, and Druitt was within a train journey from his cricket match), so they would get +1.

Because the information is in the "lower" score really, I think location scores might as well be 0 if your in London (Long isn't a suspect, so she shouldn't have a +2 suspect score), with negative values being used to indicate there is various degrees of "concern" with regards to their locations (-1 and -2, rather than +1 and 0). And if someone can definitely be placed in a location that makes it impossible for them to be JtR, then a very large negative score can be applied.

My suggestion wouldn't change any of the rankings, just lower the total. But if a witness gets +2, then really to evaluate the suspect scores what matters is how far above +2 they are, so why not just make that 0 in the first place?

- Jeff

I would alter the term generalised term of "location" to
"proximity to murder site"

And then have an added separate score for "timing" that works in conjunction with the former proximity to murder site.


In other words, the likes of...

Goldstein, Schwartz, Hutchinson, Lechmere, Cadosche, Lawrende, Paul etc...etc... should ALL score a point for being within CLOSE PROXIMITY of the murder site AND at a TIME that was relatively close to a murder having been committed.


So for example someone like Bachert who lived close to the epicentre of the murders for many years, but has no direct physical link to any of the murder sites in terms of the timing, would score 1 for proximity but 0 for timing.

Someone like Lechmere would score 2 (one for Proximity and 1 for Timing)

Dr Barnardo is an interesting one because we know for certain that he was at a conference giving a speech in Dundee within 36 hours of one of the Canonical 5 murders.
However, to get to Dundee, we also know he took a train from London.
This then confirms that Dr Barnardo left London on the morning train to Dundee on the morning of one of the murders; ergo, just a few hours after a Ripper victim was found.
Barnardo would then score 1 for proximity because he was in London for sure, but 0 for timing because he has no known link to a murder site relative to the timing.

Regarding Long; while she should technically score 2, 1 for proximity and 1 for timing; anyone with common sense would know that she has no chance of being the Ripper; not just because she was female, but because of every other factor that rules her out.

While it's tempting to introduce minus points; I would resist the temptation because having a base value of 0 works perfectly.

100% of nothing...is still nothing, despite the percentage value.

Therefore, with Deeming, there is no proof he was in London, but we know that he did visit London sometime between 1887-1889 because he accumulated debts in London...and the only way he could have done that...is if he was there in London in person.


Deeming would score 0 for proximity and 0 for timing because he has no physical link to any of the murder sites whatsoever.

However, to give minus points then by proxy assumes he wasn't in London, but considering there's no proof of that either, then he should score 0 and not be given a minus point
By allocating minus points, it subtly implies the passive ruling out a suspect that may not be deserved of that.


RD

[JeffHamm]

I would alter the term generalised term of "location" to
"proximity to murder site"

And then have an added separate score for "timing" that works in conjunction with the former proximity to murder site.


In other words, the likes of...

Goldstein, Schwartz, Hutchinson, Lechmere, Cadosche, Lawrende, Paul etc...etc... should ALL score a point for being within CLOSE PROXIMITY of the murder site AND at a TIME that was relatively close to a murder having been committed.


So for example someone like Bachert who lived close to the epicentre of the murders for many years, but has no direct physical link to any of the murder sites in terms of the timing, would score 1 for proximity but 0 for timing.

Someone like Lechmere would score 2 (one for Proximity and 1 for Timing)

Dr Barnardo is an interesting one because we know for certain that he was at a conference giving a speech in Dundee within 36 hours of one of the Canonical 5 murders.
However, to get to Dundee, we also know he took a train from London.
This then confirms that Dr Barnardo left London on the morning train to Dundee on the morning of one of the murders; ergo, just a few hours after a Ripper victim was found.
Barnardo would then score 1 for proximity because he was in London for sure, but 0 for timing because he has no known link to a murder site relative to the timing.

Regarding Long; while she should technically score 2, 1 for proximity and 1 for timing; anyone with common sense would know that she has no chance of being the Ripper; not just because she was female, but because of every other factor that rules her out.

While it's tempting to introduce minus points; I would resist the temptation because having a base value of 0 works perfectly.

100% of nothing...is still nothing, despite the percentage value.

Therefore, with Deeming, there is no proof he was in London, but we know that he did visit London sometime between 1887-1889 because he accumulated debts in London...and the only way he could have done that...is if he was there in London in person.


Deeming would score 0 for proximity and 0 for timing because he has no physical link to any of the murder sites whatsoever.

However, to give minus points then by proxy assumes he wasn't in London, but considering there's no proof of that either, then he should score 0 and not be given a minus point
By allocating minus points, it subtly implies the passive ruling out a suspect that may not be deserved of that.


RD

Hi RD,

One could refine the scoring system with regards to location along your proximity" and timing suggestions. Indeed, with any such thing one can make things more or less specific in nature. Both directions have their pro's and con's.

In the end, it is not a specific category of a code that matters overall, only the final total score. What I mean is, a suspect could get a negative location score, but fit so well in other ways end up with a higher overall assessment. Being negative on location isn't eliminating anyone simply because that score is negative (except in the case where location rules them out entirely, in which case it's an alibi, and so no matter how many points they get in other areas, they are still ruled out, which is what I was getting at with the -100 idea - the final total would be massively negative).

And the points are a relative ranking of individuals. The actual value in many ways doesn't matter, it's the relative difference. So changing location from -2 to 0 rather than 0-2 makes no difference in terms of those relative ranks.

However, I don't think one can say "Long gets 0 anyway because nobody thinks she's the Ripper" isn't how an objective coding scheme should work. Her location fits the coding scheme, and as currently implemented, that means her "score" is +2. In fact, every witness, PC, and people living in the are would all get +2 simply because they live in London. As such, that +2 is to me, pretty meaningless, so it might as well be 0. I'm not sure even coding things like "proximity in time and place" is great, as that would get applied to all witnesses, PC's, and those who live in the buildings nearby, etc. However, when evaluating a suspect, what does often get mentioned as something that makes them a weaker suspect is when their location is outside of London (or if their location is unknown). In other words, location tends to come up as something that detracts from a suspect, which to me would be reflected in applying a negative score, reducing the strength of the overall "fit".

I do get that "close in time and space" is tempting, and it is often argued as a point against, but as I see coding schemes like this as something that only has utility if you apply it everyone, then it starts to positively code witnesses and so forth, which I'm not sure is helpful.

See, one thing that I think would be useful information to consider is to apply the coding scheme to people we're pretty sure are not JtR (like Long) as then we would get some sort of idea as to what the range of scores is for innocent people. If a suspect has a final total that is pretty commonly obtained by people who are innocent, then that would tell us that suspect's fit is easily achieved by chance. But those who score much higher than is typical, could then be viewed as stronger. And with that in mind, I see it as pretty uninformative that someone is in the Whitechapel area, even at the right time, because that applies to a very large number of people.

In the end, it really doesn't make a difference because whether the score for the general population hovers around 0, or it hovers around +2, due to the value of the location score doesn't really matter. Someone with a score of +6 or +8, is still 6 points higher than the general population would get, and that is the important information.

And it's important to remember, a negative score on location is another way of indicating that location is a weakness in the case against this person, something that I think gets lost in the 0-2 coding. But that's just me, and as I say, all it does is shift everybody's total down by 2 points, it doesn't change anything about the relative rankings, or the difference in rankings between two suspects. It's similar to converting inches to centimetres, nothing about the actual length has changed, only the numerical value; but I'm just suggesting subtracting a constant here and there's no multiplying by fractions to rescale as well, so it's a bit less of a transformation than imperial to metric measures).

Seriously, all one would do is take the current totals, subtract 2 from everyone, and after that, interpret things exactly the same way. I'm just being a bit pedantic about measurement (partly because I like the idea overall, and to me location information is really non-informative because of how many people would score the current +2, and what is impactful on a particular suspect's score is when they get less than the +2, which to me implies the information is contained in the reduction, hence it operates like a negative value, so why not make it one?)

- Jeff

[Lewis C]

Hi Lewis,

Yes, that is exactly my point, but you said it much clearer than I did! Thanks!

- Jeff