jordanchan

Yeah, I was a bit disappointed to see Donald Trump dragged into this, almost arbitrarily, and almost needlessly. Then I remembered I was afterall reading the Vox.

Of course it isn't new. Very little is truly new. It is the same old stuff repeating in different contexts, leading us to comfortably believe "yeah, but this is different".

Have we already forgotten of the ilk that harassed and harangued Galileo and Copernicus? Even Socrates?

This is especially applicable to the scientific cadre; sometimes leads me to think that some scientists are just as close-minded as the religionists of yore. The same applies to "believing in science". If I have to believe in it, it is no longer science. Conversely, by its very nature, the scientific approach must be open to invalidating its own self.

I often wonder if the Jewish tradition of casually and constantly debating all manner of things is a good antidote to getting rigidly attached to one's own viewpoint. But on the flipside, nonstop argumentations are a nuisance too.

You're right that there are few rewards to humility.

An example that immediately comes to mind is during my thesis, I set upon the task of creating a multivariate model to predict an observable quantity. My advisor kept insisting that my goal was NOT to build the model, but to first see that IF the variables had enough predictive power. I just never understood (at the time) what he meant. I went straight ahead and hammered out a robust model, and got a good grade (reward).

The rewards are intrinsic, not extrinsic. Sometimes the rewards can be extrinsic in the long run - if you find and invalidate a hidden assumption and are then able to build a theory/product/business around it. But in the short term, the rewards, and hence motivation, is usually intrinsic at best. Often there are no short term rewards, because you're in a perennial state of discomfiture owning to questioning everything.

The US Gov has a very nice little book on the topic - A Tradecraft Primer: Structured Analytic Techniques for Improving Intelligence Analysis. Just google and you'll find it on the CIA's website.

The book is geared towards the intelligence community but can be appreciated by all intelligent communities, especially on a topic such as this *. In a nutshell, it is about discovering implicit assumptions, questioning them, and checking for alternative explanations - in all areas of analysis/thought.

The idea of selling to people based on their needs assumes that everyone has needs/insecurities which, if correctly targeted, will result in a successful sale.

To this extent, we're fully aligned. Take a step back. To the theory of needs, one of the earliest and best known variants of which was proposed by Maslow. (There have been numerous updates to Maslow over the years [1]. A common theme in all variants of the theory of motivation/needs is the concept of self-actualization.)

If individuals' needs are the foundation of selling to them, someone motivated by love and belongingness needs ought to be motivated by things that will help him feel closer to others. Indeed, this is the base of the couple of examples we discussed (pretty girl selling snake oil, and the supermarket with good looking sales girls). Similar arguments can easily be put forward for each and every kind of need - doesn't matter what the actual product is, what matters is what the buyer thinks/believes it will help him achieve, ergo, what need(s) will it is perceived to fulfill.

The crux of my counter-argument is based on the exceptions - the self-actualized individuals, who admittedly form a very small fraction of the population. You're right in the general case, of course. I'm saying there is a proper subset which is an exception to the rule. Going by the theories of motivation/needs, the self actualized people are driven solely by what they believe their "purpose" in life to be. It is also supposed, in these theories, that self actualized individuals either do not have needs (what you rightly phrased as insecurities), or aren't driven/motivated by them.

In the general case: As long as there is an illusion that need x will be met by a product y, people who are motivated by x can be sold y. In the case of self-actualized individuals, it is harder to sell an illusion - which won't be sustained in the face a logical/rational analysis. Since there is no underlying need/insecurity to target, all that remains is a factual analysis of what material benefits the product confers vis-a-vis what the self-actualized individual's goals are.

So I guess it boils down to: what's your take on the theories of motivation and the notion of self-actualized individuals?

[1] I found a good summary + a new approach at a Medium blog post, which seems reasonably well researched and presented. https://medium.com/@faracrosstherubicon/rethinking-maslow-a-...

They just started YC research. Maybe its time for YC Movies...

Hydroponics using atmospheric humidity maybe?

Reading this made me wonder...do SpaceX and other private space flight programs have similar standards of rigor?

Agreed - everyone has their weak spot, but not for everything. Its like everyone has their price, but not the fellow considerably richer than you.

As for smoking, if companies were allowed to actually misstate it is good for you, there would be myriad other organizations working to prove it isn't. A reasonable person might well do his own/independent fact finding.

Someone can say eating an octopus is yucky [it is actually used with salad in many places] but for a large amount of money you can convince/bribe them to try it. But if Warren Buffet says no, good luck bribing him. You can try another [practical, not theoretical] incentive, but I'm sure he could buy that too. One can say rich people must live certain lifestyles, but Steve Jobs lived an almost monastic life [save the Mercedes and a couple of other things maybe].

That inherent need has to be present for an advertiser to target/exploit it. And the person has to actually buy the premise of the proposition. Many have the guts to go against the crowd and some even take it as a badge of honor.

Valentine's day is a manufactured custom too. Many are religious about it, many others abhor it - despite the concerted, and mostly successful, efforts to create a romantic custom out of the day. Diamonds, I agree - in many cultures, especially the US. But in many other cultures, such as the Nordic, it is less common [1 - a forum thread]. That doesn't mean de Beers doesn't try its level best to manufacture the same kind of social custom in Sweden. That need for extravagantly showing off a romantic pre-commitment just doesn't exist in that market.

Now let's reasonably assume your SO is well aware of the hiring-pretty-girls-policy, and does not prefer you buying from that store given that there are better economic alternatives, and add in a touch of reasonable jealousy. Do you still frequent that store, given that you are happy with your SO?

Could I sell you an extravagant snake-oil panacea if I were a hot bird?

[1] http://www.thelocal.se/discuss/index.php?showtopic=32620

Indeed. Everyone has their sensitivities, vulnerabilities and emotional/other needs. A company that has figured out how to push the right buttons in its sales pitch/ad has found its subset of "gullible people" - who will be driven more by animal passions than by reason. Humans are very well capable of overlooking potential long term harm in favor of short term pleasure, or even convincing themselves that the ill effects won't apply to them.

That being said, I wouldn't agree that every single human being is gullible when attacked from the right angle. Example: there are many stressed out people who don't smoke, it doesn't matter how or how much you try to convince them. One can peddle snake-oil on the streets, but only a subset of passersby will stop by and an even smaller subset will buy it - this does not imply that only the most afflicted are the ones making the purchase.

Anything can make fools of gullible people. Free markets or oligarchies.

There could well be discriminatory billing - clean conscientious passengers pay less than drunk party goers.

Cities where Uber and the like are being blocked with oppose robo-cars even more vigorously. Like the revolts against steam engines and factories we read about. Uber-opposition will seem like a practice exercise.

On the supply side: No patents, or easy/cheap to license patents. Technical trade secrets don't remain so for very long. Large margins for one or a few suppliers/manufacturers.

This is basically what is happening with the smartphone business. First Apple started it. Then there were a few players like LG, Samsung, HTC, etc. Now everything and their uncle is starting a smartphone company. Companies like Lenovo, video con, not to mention the Asian upstarts from China and India which are definitely late entrants to the market.

On the demand side: Steadily increasing users: Early adopters not abandoning the product, but continuing to use it. Increasing numbers of early adopters in various social circles.

I had tried this a while back, and found it depends on what you're searching.

For technical topics, I had found bing had a preference towards the articles from the microsoft developer network. It also had a bias towards MS technologies.

For example, looking for something vague like real time web server, bing gave a couple results from asp.net pages on the front page, while google seemed more balanced. Google had one link to NodeJS while bing had none.

You could experiment this yourself. I sometimes keep different tabs with different search engines and pick what I like best.

This is a good topic, and dear to me too.

First and foremost requirement: people with some experience in the various areas of learning. Ideally, not professors; professional educators can have biases this doesn't need.

If it is to catch on in any form, it needs to offer significant tangible benefits over current methods.

0. Lessons need to be byte-sized. If someone wants a lengthy single-session lesson, they can combine smaller segments. But the byte-sized segments should be complete in their own right.

The overarching format should be tree-like, with the technical depth and sophistication increasing as one goes down deeper in the tree. The point being, someone should have the ability to take a few lessons, and still come out reasonably educated about the topic. Instead of, say having to take an entire course to get the full picture - which is the case with most current education programs.

Social => interaction => Q&A. So, ideally, things should have a Q&A format. This can be preceded by an intro to the topic exploring the scope of the subject, and explaining how the individual lessons fit into the learning plan. The other advantage of a Q&A format is hands-on learning.

Interaction also implies demonstrations, exercises, experiments/simulations, and yes: animations.

1. It needs a community of tutors - this can be sourced from the various stack exchange sites, and from Quora too. It is important to ensure that there are at least two different versions of each lesson, to offer varying perspectives, or at least a different style of presentation to suit individual learners.

2. Needs a volume of learning material. This will obviously take time to build up, and is a classic chicken and egg problem. Do the students [i.e. demand] come first or do the lessons [i.e. supply] get compiled first?

A good hack might be to take current well-received answers on some of the aforementioned sites and use them to compile individual mini-lessons. This will allow to sample the demand and give directions on what subjects/topics need to be focused on.

Additionally, video content format needs to be created - this can be people talking, or ideally, demonstrations, animations, etc.

3. It could be for a) hobby-learners, or b) oriented towards specific programs, like a certification. It can be a good idea to start with a and move on to b as/if things pick up. Once things have moved to b, it could go seeking for independent recognition - in a manner comparable to current certifications. Especially, recognition from employers is important to validate this as something people can bet time and money on.

Alternatively, the whole thing could latch on to one of the existing MOOCs, but that comes with its own set of challenges. But it might be better to use those as supplementary material, i.e. the backend, while this becomes the front end. The third possibility is this becomes a supplement to current college courses - the way students presently use the various q&a sites, but at a higher level, in that there would be original and organized content on this. Eventually, some progressive educators will start referring to this to design lesson plans.

4. The core of social: there have to be points/ratings, and eventually a way of translating from those points to money. People get points for good explanations. The points system could be inspired by stack exchange, quora, and most importantly: the slashdot voting system. The last one is important, because it is not just simple upvotes and down votes, but allows students to vote lessons on different categories - insightful, informative, etc. This is vital for lessons, in addition to an overall rating system.

This is just something off the top of my head. This area is a significant challenge - and potentially promising, that I have not yet delved into. Are you actually building something?

feedback/comments are appreciated

Thanks for sharing this! I would love to help with this - once I have gotten up to speed on various things by building a graph on Postgres myself.

The point about micro services is a really valid one. In principle, if the application is neatly split into a group of micro services, it would make things easier. This is something I am going to look into once the application has evolved somewhat - it is hard to make the split from the get-go, without knowing how it is going to be actually used - by the end user.

Yes, it might be a hack, but it should offer some powerful functionalities... I'll remember this one.

Yes, it might be a hack [actually I don't think it is], but it should offer some powerful functionalities...

Thank you so much for this list. I am looking into these... Maybe there's something I can reuse, but at the very least I can get some inspiration from checking them out.

AFAIK:

A proper graph db is almost mandatory if there are complex, ad hoc queries that need to be made on real time data.

Using an RDBMS is great if the graph query types are going to be known in advance - so they can be prepared for using materialized views and indexes, and if they aren't too complex - so one doesn't descend into JOIN-hell.

But most applications aren't like that, and not all applications can be completely satisfied using only a graphdb. Hence the rise of the new multimodel databases like OrientDB and ArangoDB. So I think it is a question of what risk one is prepared to tolerate.

Thanks for the links. Glad to know there are more experienced people pursuing this seriously!

This made my day! Thanks, I will check it!

I noticed the same - the "explosion of tables" when I was doing the actual designing. But, almost accidentally, it also turned out that the tables were almost all either 5NF or 6NF. So I thought this is just a highly normalized setup anyway - something traditional databases are supposed to be designed for.

The list of Postgres-basd extensions mentioned in this thread turned out quite relevant actually! I'm still going through them all, seeing if I can reuse something..

Many thanks for the inputs, quite helpful.

Thanks, that was almost the same reasoning I used to settle on this approach.

I definitely need acid transactions, and also some graph capability - which may or may not grow in the future. More pertinently, the kind of representation I was attempting led to a very normalized design anyway, and I think it also lends itself to all but the most sophisticated of graph functionalities! But I am only just starting, so hopefully it works out well.

You raise a good point. I don't expect graph queries that go more than a few links away. To the extent that there will be such queries, I think appropriate indexing will help a great deal.

Implementation wise, it is not that different from a query with many joins anyway - which is a standard problem, and is addressed by materialized views as well as indexing. So as long as the complex queries don't need to be ad hoc and real time, I can prepare for them using materialized views.

However you are quite right in that complex, ad hoc, and realtime graph queries Will be a pain point. That is a risk I am prepared to tolerate in the short term. The thing is, I also need transactional integrity, so I am stuck with an rdbms as the primary solution. In the long term, I anticipate the need for both graph queries and transactions - thus, I think it is better to start with postgres and incorporate a graphdb as the need arises than the other way round.

Hopefully this line of reasoning make sense? I am really not experienced at this, so am trying to not bump into too many walls.

True, as I was designing the tables, I was surprised to notice how normalized it looked.

The generic kind of queries you mention: actually, yes, I think I will need them.

Your second point on downsides is something I have/had to think about, thanks for raising it. I had some vague premonitions on those lines, but you helped make it concrete. The problem can be avoided by not making meaningless connections, but that's not a real solution. It won't be a preventive, but it could help to audit relationships - by having a script list out which tables FKs originate from.

Since I started writing this post, I have had two ideas on how to address this, it'd be great to have your opinion on them - I'm thinking out loud here.

1. Use referential integrity. Since each relationship/edge will be in its own inherited table, one could impose a constraint that each column_in_the_specific_edge_table REFERENCES another_column_in_a_specific_node_table. Like the columns in the person_lives_in_place relationship table must reference columns in the person table and the place table - each of which is also its own inherited node table.

2. More convoluted and a bit cumbersome, but if the previous/simpler approach is insufficient, one could create data types corresponding to each node type. And impose that as a type constraint on the edge tables.

But maybe the 1st option could actually work...what do you think?