Peter Nyikos wrote:
> On Friday, March 22, 2019 at 11:35:03 AM UTC-4, Burkhard wrote:
>> Peter Nyikos wrote:
>>> On Wednesday, March 20, 2019 at 5:10:03 PM UTC-4, Burkhard wrote:
>>>> Peter Nyikos wrote:
>>>>> On Sunday, March 17, 2019 at 1:15:03 PM UTC-4, Burkhard wrote:
>>>
>>>>>> I'm currently part of three STEM-lead research projects, where computer
>>>>>> scientists, mathematicians, anthropologists, psychologists, sociologists
>>>>>> and media study folks work together on better and safer products.
>>>
>>> The adage "too many cooks spoil the broth" comes to mind here.
>>
>> based on your extensive experience in software design?
>
> A well known adage of the 1960's is "It doesn't take a weatherman to
> tell which way the wind is blowing."
Quite. Did I ever tell you that in my considerate opinion, we should
close down all mathematics departments, because don't you know, all the
young ones these days have pocket calculators and similar gewgaws, so
there really is no point teaching that stuff any longer?
>
>
>> Also
>>> "How many ______________ does it take to screw in a lightbulb?"
>>>
>>> Now that may be unfair, but if so, I would like to know how all
>>> these specialists are needed for *products* as opposed to *policies*.
>>
>> Well, to give a few examples, under the GDPR, all products (which
>> includes software products) that process personal data must by law have
>> "data protection by design". This includes also data security. We know
>> that technologically unsophisticated users are the main security risk
>> (not updating software, disabling firewall etc) and also are quite bad
>> at doing risk assessments. So for law compliant software that keeps
>> people safe, especially in an IoT context, you need "usable privacy",
>> things that protect users in real life.
>
> From what? you don't seem to explain below.
Since the topic as I said is "data protection by design", I thought that
was obvious - misuse of their data, either because they unwisely shared
it, or because the security of the system was compromised.
>
>> For this you need to understand
>> why people make no or not sufficient use of some existing tools. That's
>> were cognitive science and psychology comes in.
>
> "why" is not relevant; it is, "what can we do to take care of such people"?
And to do this efficiently, you think we don't need to understand why
they make the choices they make or engage in otherwise risky behavior?
If you were to design a better rear mirror for cars because evidence
shows that drivers don't look sufficiently into it, don't you think it
matters if they do so because the angle of current models is awkward,
their size is too small, it interferes with the other displays they try
to look at at the same time or everything is fine with the mirror, they
just lack the training to use them right? Now I would say that depending
on the answer, you'd chose very different strategies to improve the
design. Similarly, if I want to increase uptake of a beneficial drug, I
might want to know first if people don't use it because it's too
expensive, because they suffer embarrassing side effects, because they
are too big to swallow comfortably, they interact with other medication
they have to take etc etc So before I can start to think how to improve
the drug, I need to know why people don't use what we have.
>
> And the answer all too often is, give the design to people well versed
> in advertising. It used to take only two clicks from the university webpage
> to get to the Department of Mathematics webpage. Now it takes twice that many,
> and the relevant "button" is not always easy to find. You have to search
> for the one for the College of Arts and Sciences. But when you get there,
> to a page dominated by advertising, the right button asks you, in effect,
> "Do you really want to look up the College of Arts and Sciences? then click here."
> The page to which the click takes you is almost all advertising that
> extols the College. Hidden in one of the paragraphs is a link -- not blue,
> but garnet, the university color -- that takes you to "Departments and schools"
> and there is where you finally get to click for the Math department.
> then finally enables you to click for the math department itself.
Now, that could be bad website design. It could however also be a
response to changing user habits - people googling for info and ten
going from the google website rather than the university launch page.
That's why most web developers would use something like google analytics
to find out how people get to the info they want, how many false tries
they make etc etc.
Now that is too basic to be part of academic research, that's normal
design procedure. You could ask if it is not unfair to older users like
yourself who have to adapt their search strategies and might as a result
feel excluded. That would be closer to what we sometimes do. But still
not enough meat to it. But what is an interesting research question that
one could tackle in a small project with computer science and other
relevant experts from the social sciences could be if a) more generally,
changes of design in response to changing user habits excludes, rather
then merely inconveniences, older users (which could be an issue for
websites that are necessary to navigate, e.g. by government agencies)
and if so, if there are design solutions for this that make transitions
more gradual.
>
> Big universities are run like big businesses, with Administration in the role
> of management, faculty in the role of white-collar labor that often has to
> defer to real blue-collar workers, and students in the form of customers.
> Hence all the advertising in the brave new world of search software.
>
>> These systems are
>> physically embedded, so users interact with them through bodily
>> movements etc (think of IoT-enabled jewelry e.g.)
>
> And problems with sensitivity to the materials, that can take a long
> time to show up. Good luck on having it show up right where the implant is.
Not talking about implants here (though they would be one example),
merely IoT or "cyberphysical systems". So you go into a lecture room and
use hand movements to activate microphones and screens rather than
pushing buttons or voice commands. Has lots of applications, e.g. if you
work in noisy environments, or where lots of people talk at the same
time (i.e. industrial manufacturing environments with robotic
co-workers), environments were silence is needed, or to assist people
who can't speak.
>
>> To understand how a
>> user could manipulate them in a secure way therefore needs also
>> knowledge of ergonomics and generally human movement and interaction.
>> These then also have to work in social contexts, i.e. depending on the
>> choices other people make (herd immunity is needed), so one also has to
>> understand how thinking about privacy and data security happens in
>> groups ("all my friends do X, therefore it's probably safe" etc) And all
>> these insights then need to be turned into design solutions, for which
>> you have the artists and media study folks, where one can use e.g.
>> various forms of visualization to alert people in non-intrusive yet
>> effective ways of risks and what to do about it.
>
> Sorry, I have no desire to become a cyborg.
So no peacemaker for you should you suffer a heart attack? No
intelligent cochlear implant if you lose hearing?
Whose idea was this tampering
> with bodies just to get some sort of unspecified security?
The "tampering with bodies" was really your idea, not sure what you
mean. But as far as security and tampering with bodies is concerned,
that would be military applications, mostly. So that e.g. the commander
of a unit can get real time reading of the vitals of the personnel under
their command, and can send help when a problem is identified. There are
also safety critical civilian applications, e.g. monitoring pilots, or
during a humanitarian crisis keeping track of people.
>
> Did anyone with expertise in philosophy not suggest that the human race
> might be better off without such products?
So now you are saying we did not involve enough different disciplines,
sort of the exact opposite of what you've argued so far? And yes, the
philosophical issues are part of the "ethics and law" remit. And as so
often, in real life they turn out to be more nuanced and difficult than
what the man in the pub, or the non-weatherman predicting the weather,
might think of in 5 minutes. Why, one might even think that it requires
careful research and expertise, some of which one might want to give to
students while undergraduates. So in a STEAM curriculum, the computer
scientists who ten to be ever slightly to enthusiastic about just
building stuff get a bit of exposure to things like "cyborgs in
literature", fro the Golem of Prague to Pygmalion or Karel Čapek to get
a bit sensitized to the idea that there is something that goes to the
heart of human self-identity when we turn ourselves into machines. Or
some courses in anthropology based ethical systems, where I for one
would happily include a discussion of "Laudato si"(and have indeed
someone from divinity lead one of my classes) , together with
anthropologically based ethics by other religious and non-religious
traditions,
But all that according to the author of the paper would be diluting the
objectivism of science which neither wants nor needs this type of
detraction from the things that really matter.
>
>>
>> In another project, the outcome will be a product that assists border
>> police to identify smuggled art. That means it must be able to
>> distinguish between a £5 replica from the museum shop and a real
>> thing.
>
> So far, so good, but what's the point of software replacing a human
> policeman to make the following judgments?
The short answer is that it's terribly difficult to make people with
advanced degrees in art history join airport security, where they work
an extremely boring job just above minimum wage and get insulted by
irate members of the public a lot.
The longer answer needs a discussion of training and career structure in
the police, self-understanding of police/border security officers and
how they see their role, understanding of people flows in airports (and
how much delay per search event would make the system crash - short,
lots and lots of questions for which relevant expertise is available in
the social sciences, especially criminology. Hence multi-disciplinary
studies. The one you have issues with due to a cooking analogy, remember?
>
>> To do its job, it needs to know what the substantive law says
>> (what art is protected) and also what the procedural law says (what type
>> of evidence is needed, how does it have to be stored and verified).
>
>
>> And
>> then it needs to know a lot about art history, and the type of red flags
>> an art historian would look out for when making such an assessment. And
>> it then has to present all this in a way that makes sense for the
>> intended user,but is also legally waterproof.And this requires in many
>> jurisdictions to quantify the accuracy of the assessment, which is where
>> the forensic statisticians come in
>>
>> And we also work with artists who try to use the carbon footprint and
>> ethical sourcing of their products as a unique selling point. To make
>> this a convincing case to their customers, and also to comply with
>> advertising law, they again need a way to integrate this information so
>> that customers can easily access and understand it. That again involves
>> various data visualization approaches, but because the product is art,
>> here they should be informative, reliable and also beautiful.
>
> Yeah, like the art on all those webpages en route to the math department.
Is your department selling artwork? Now one of the artists in residence
we are working with works with silver and gold. One of the ideas we are
exploring with her is an extension of an older project on "Tales Of
Things and electronic Memory." So an engagement ring e.g. that
"remembers" the wedding and the honeymoon, and maybe key moments of the
shared life afterwards. So if the grandkids come to visit, you point it
at a smartphone and they can see how the world was like when
grandparents met, how they looked like, the first moment the ring was
put on the finger etc etc.
Is this a life's necessity? Arguably not, though there are some
interesting uses to assist people with dementia using the same approach.
But it raises a lot of technical, ethical, legal and philosophical
issues if you want to do it right, enough to argue that at the moment
they go beyond what companies do on their own in-house anyway, and
requires basic research together with industry partners. And if that
creates also a few jobs in Scotland then I'm all for it
But to cut this short - I know this might look like a radical idea, but
do you think that maybe, just maybe
- if five senior academics draw up a plan for a major research project
- it passes four levels of university-internal peer review (three at the
collaborating schools, one at cross-university level, all stacked with
academics working in the field)
- passes three levels of external peer review and final approval by the
national science funding body, so that also 17 other senior academics
looked at it)
- they manage to convince industry partners (from computer manufacturers
to big banks to festival organizers) to provide matching funding to
the tune of £6m
that someone, somewhere would have spotted if there were something
obvious wrong or irrelevant in the approach, something of the type an
outsider might spot in 5 min, or that maybe, just maybe they bloody
know what they are doing, including relevant research questions that
haven't been asked or answered yet, and the right type of mix of subject
expertise?
You know, just, as a radical thought?
>
> And once a student, and parents, decides USC is good for them, the
> wooing ends, and the honeymoon ends when it is revealed that freshman
> have to live on campus unless their family lives in either Richland
> or Lexington counties. They also have to take the dorm room that is
> assigned to them, even if it is a musty basement room, or has a thermostat
> over which they have no control, along with a sealed window. Up in Clemson
> they can't even control how fast their fan run
>
>> putting e.g. a QR code on them would not work. So you need to be able to
>> think also like an artists, and integrate aesthetics, (consumer)
>> psychology, (design) informatics, a bit of legal oversight and also a
>> bit of material science at the integration stage.
>
> Oh, and a bit closer to home for you: Oxford victims of sexual coercion by
> boyfriends are discouraged from reporting it. Bad PR for Oxford, and all that.
Not quite sure what point you want to make here?
>
>
>>>>>> Almost
>>>>>> all of the lead investigators are dualy qualified, but we struggled to
>>>>>> get also enough high quality postdocs that combine STEM with humanities
>>>>>> - that's also because industry has spotted how much additional value
>>>>>> they bring, and they can by now pretty much write their own salary
>>>>>> cheque.
>>>
>>> I addressed this in my first reply to this long post of yours, but kept
>>> it in for context. Here too:
>>>
>>>> But everyone who knows that field knows that it is big companies that
>>>> are desperately looking for dual qualified people, and willing to pay
>>>> for it,
>>>
>>> The cynic in me suggests that old fashioned nepotism might be at work
>>> here, with companies very carefully tailoring the job description
>>> around the person they have decided to hire, and only advertise it
>>> because of legal requirements.
>>
>> Ad you can back this up with evidence, of course? Not just anecdotes?
>
> Yes. See below about the two cases where I was intimately involved.
Two cases pretty much = anecdotes, don't you think?
>
>>
>>>
>>> I've seen that sort of thing happen, to people I knew and cared very
>>> much about, on both sides of the deal. One friend got a job due to such
>>> careful tailoring,
>
> For one thing, she was a minority temporary faculty member, so part of the
> description was being able to counsel students in that minority. I know,
> because I was asked to write a letter of recommendation for her. I wrote that
> I really had no firsthand knowledge of how good a counselor she was, but
> I knew she was a member of the coveted minority. Otherwise, stuck to her mathematical
> research and scholarship.
Not quite sure what conclusions you draw from this. Did you say in the
reference her scholarship was abysmal? Then and only then do you have
(some) evidence that it was tailoring that got her the job. If not, how
do you know that she did not get the job because of her strength in
research that you testified to? Indeed, if "tailoring was involved,
candidates with a referee who did speak to their counseling ability
would have had the upper hand. So if anything that would indicate that
this did no play a decisive role.
Second, academic roles are more than just research, for most of us, and
the university can only function well if these too are done
professionally end expertly. Very often, this does not happen, and these
things are treated as an after thought. So I was e.g. for many years the
disability officer of our faculty, without much in terms of prior
training and expertise. A job that if you do it badly, life are at
stake, so if anything something where relevant expertise would be more
relevant than even in my core research field. Not being disabled myself,
and with no relevant education background, I tried to do my best, learn
as fast as possible and listen to as many students as I could - but
without a shadow of a doubt someone with direct personal experience
could have done a better job.
So I would have found it perfectly OK at the pot of application, if the
university decided it had to get its act together to assist students
with disability, to use relevant experience at the every least as a tie
breaker between me and another candidate.
>
>
>> the other lost out on a (completely different) job whose description she fit perfectly. Had I learned about it soon after it had
>>> happened, I would have advised her to look into suing the company for
>>> discrimination.
>>
>> and you know the qualities that the successful candidate had too?
>
> They included extra ones that were not in the job description. The unsuccessful
> applicant was asked about some of them in the phone interview. She met them
> very well indeed. When she asked why she didn't get the job, the employer
> made the mistake of saying, "the one who got the job had music experience."
> Reaction: "Why didn't you ask me about that? I could have told you that
> I have extensive music experience too!" [Eight years of piano lessons
> and one year of playing French Horn in a school band.]
>
> You can bet that employer never made that kind of mistake again! There's
> the dark side of adding A to STEM.
And how would that differ from adding competency in a
language/experience of studying abroad? Or a specialism within the
subject, or a different Stem field ("yes, your math is good, but the
other candidate has experience in biology, and as part of the research
is modelling of population dynamics, we took him")
Now, without further detail it's difficult to analyze this case -
"could" be a massive and obvious violation of HR practice, which indeed
could give rise to legal action.
But here a variation of the story. Say the company develops games, and
needs a mathematician to build more credible AIs ans Non-player
characters (something that involves lot of game and decision theory,
hence they do hire also from our math department). In their CV,
candidate 1 mentions that they were member of an electric band, and that
he has lots of experience writing software code for electronic music and
embed it in videos. In the "what do you bring to this job" session, the
candidate says that having played the games of the company, he thought
the experience could be much enhanced though communal chanting before
the battle, and they already have some ideas how that could be done.
Nobody in the company had anticipated this in the job description. But
it showed the candidate a) had thought carefully about what they want to
do in their job and b) brings an interesting new skill, which separated
them from the others.
So this becomes a short "the other guy did music" in the post-interview
feedback, often given by someone junior who acted as a mere notetaker
and did not understand why that was significant. Now, would they go back
to all the other candidates and ask them: Oh, by the way, you did not
mention this in your CV or the interview, but we found one candidate who
also does interesting things with music, could you do this too? Not
any appointment panel I ever sat on, and I did a fair few. There is in
most interesting (i.e. most graduate) jobs a degree of unpredictability,
where different candidates may or may not be able to use their personal
experience to argue their case.
Can this degree of professional judgement making be abused? Of course,
and sometimes it is. If the problem becomes systematic, someone should
use some math to analyze the patterns - but that would be the kind of
math for social justice that the author of the paper incompetently tries
to ridicule.
>
> Are you quite sure this kind of thing doesn't take place right under
> your nose?
>
Course it might. But a) it has nothing to do with STEAM in particular,
you could as well use a different Stem experience, or any experience,
from "traveling abroad" to "job experience as tour guide". B) there is
no evidence offered that moving to STEAM curricula makes it worse - on
the contrary, one could argue that once everyone had some "A" exposure
while doing STEM, it becomes less suitable to differentiate between
candidates. And third, to combat this the very disciplines that are most
focused in looking at and analyzing systematic abuse of power in
situations of power imbalance, such as hiring decisions, are exactly the
type of subjects the author attacks.
So in summary, you offer to anecdotes, which remain inconclusive even on
their own terms, and even if they were examples of "tailoring" would
not substantiate the claim that a STEAM curriculum makes them worse, and
indeed some evidence to suggest that the opposite would be the case.