Lost Week: It's Like Shark Week, Except Different in Every Conceivable Way

Last Sunday night, Lost ended with a two and a half hour extravaganza. This sparked a fury of conversation between Joe and I which was interrupted by the demands of real life. Neither of us were content to let it end with that, and so we have decided to let the dialogue seep on for awhile. As there is a real void of Lost related discussion on the internet by random people, we thought we would chronicle this discussion online on this here blog. Anyway, be forewarned that we will definitely spoil many major plot points for those of you haven't seen the show and plan on doing so.

The following entry was written by Joe.

In the words of Dewey Cox “there’s something happening here, what it is ain’t exactly…. obvious”.  That quote correlates well to what I’ve taken away from the show, the finale, what the island was and what the show was really all about. The more I think about, the more I find myself disappointed with the writers’ chosen approach to conclude this amazing and historic show.  

As the urban idiom goes, “haters gonna hate”, but even haters must concede that Lost was an unprecedented (and likely to be unequaled) achievement of network television. It was the antithesis of most network goals in a sea of compartmentalized dramas and cheaply produced reality television. It was expensive, complicated, genre focused and exceptionally serialized to the point where each episode had to be watched in order, and along with 24, really kicked off the TV-on-DVD revolution. A show who’s driving focus ab initio was the finale. This was a show that achieved things in a way that Twin Peaks teased at without the benefit of an endgame (or the internet), and the X-Files flirted with but never embraced the hidden-clues-coming-together-at-the-end-you-have-to-watch-from-ep1-to-really-appreciate kind of storytelling Lost did. It was without hyperbole one of the best examples of what network television can be.

Before I get into things however, some general comments I believe are in order, before we delve into the nitty gritty of specifics throughout this week. For those who don’t know (which is everyone save for 3 people) I will be providing a divergent viewpoint than that of Rick throughout our discourse. I don’t imply viewpoint as in my opinion of the show/finale differs from his (though I suspect it might) but divergent in that I literally viewed the show from a different position than Rick. Not to suggested an air of authenticity or entitlement but I was “hardcore” as the kids say. Starting with the 3rd season four years ago, I was forced to wait a week between my episodes, and then months between seasons. These were incubation periods where the desire to find answers to the shows mysteries multiplied to critical conditions. Cliché though it may be to say, Lost wasn’t just “a show”, there was an entire sub-culture underneath it that the producers actively propagated that communicated to the fans that the island’s mysteries were integral to the show (more on this later). There were dozens of “podcasts” (like radio shows on very specific topics) that tossed theories back and forth and re-caped episodes. There were online interactive stories to occupy hardcore fans during the frequent hiatuses that provided more mysteries to the island. Take a look at some of them at LostPedia and understand how (perhaps other than Star Trek-dom though it lacks the overarching mystery) Lost became a part of audiences lives in a way most shows could never compare. And it is from this viewpoint that I can say the finale of Lost both left me sad to see it go and also significantly disappointed.  

Not disappointed with the final scenes of Jack stumbling to his origin point, or even the non-corporeal dénouement; in actuality I really enjoyed the finale 10 minutes of Lost and think they would have been perfect conclusions if there was more exposition throughout the 6th and final season. Lost was a show that frequently and consciously kept the viewer returning with a tantalizing world of mini-mysteries that the show runners promised time and time again would pay off. We were encouraged to be Lockian men of faith in contrast to Shepherdian viewers who walked away from the proverbial hatch button and assume the show couldn’t deliver on its promises. It’s easy for me to envision that JJ Abrams pitched this show and planned out the first episode and truly did have the foresight to say “and then the show will end with the flashbacks of the final season turning out to be a kind of purgatory with all these season 1 characters and the show ends with a plane taking off with whomever you choose to keep alive and Jack’s closing eye”. I’ll give them that, I think they had that in the bag all along, and it is quite a good conclusion. The problem that I’m beginning to realize is after that point, when JJ handed the show off to Damon Lindeloff and Carlton Cuse, they didn’t understand how to get there properly. They literally built mysteries on top of mysteries to keep viewers coming back only to go into the final season treading water to build to an interesting season ending twist and then said “the show is about the characters more than anything else”. To that I say horseshit (pardon my German). This show made its living on the promises of answering the many mysteries in a satisfying way. It is not simply a “character show” like Grey’s Anatomy or most television shows (not that there’s anything wrong with that).   A show like Friends is a character show, where if one is vaguely familiar with the show they might ask a fan “so who does Rachel end up with in the end?” As a Lost fan who works at a video store I’ve heard a myriad of questions from casual viewers and not one was “So, who does Kate end up with?” or “Will Sun and Jin ever be reunited?” I always hear “do they ever say what the island is?” or “what is the smoke monster?” or “why are there polar bears?”  These questions literally are the show. To abandon the mysteries in the final hour is an incredible disservice to the show even if the character conclusions were flawless.

To conclude my opening statements I put forward the following thesis: the conclusion of Lost provided exceptional, virtually perfect emotional closure to the series but hardly any narrative closure; as a show that built its longevity upon the foreshadowed  unknown being satisfyingly resolved it can only be described as, at best, disappointing.

Now, let’s begin.

A look back at Lost by some random over-educated dudes on the internet

I haven't had cable for a few years now, and even when I did, I didn't follow too many serial dramas. However, about a year ago, a friend of mine, Master Joseph Buscemi, implored me to get in on Lost before the show ended. Even though I respect the Master's opinion on television, movies, and pop culture, I resisted for awhile. I did so mostly because I enjoy being obnoxious, but also because I worried about getting sucked into a huge time-sink. In the end, I finally agreed and I even managed to catch up on the first 5 seasons of the show before the final season began.

Last Sunday night, the series ended with a two and a half hour extravaganza. This sparked a fury of conversation between Joe and I which was interrupted by the demands of real life. Neither of us were content to let it end with that, and so we have decided to let the dialogue seep on for awhile. As there is a real void of Lost related discussion on the internet by random people, we thought we would chronicle this discussion online on this here blog. Anyway, be forewarned that we will definitely spoil many major plot points for those of you haven't seen the show and plan on doing so.



Well, it looks like I (Rick) am kicking of Lost week, but what may turn into something bigger. Here goes: there were two main threads in Season 6: the wrapping up the on-island action with Jacob and the Man in Black being the driving force (I still can't believe they refused to give this guy an actual name); and, the parallel universe. The latter turned out to be a purgatory of sorts for most of the people who were on the island (but conspicuously not all) who constructed it so that they could be together when they go off into the afterlife. At least that's what I think it was. Regardless, I suspect we can agree that it was mostly the writers' way of having their their cake and eating it too.

They needed to be able to kill off characters for dramatic purposes. However, the audience had become quite attached to a number of these people. So the writers also needed a way to be able to treat them lovingly --- to let them ride off into the sunset and live happily ever after. Well, not live, per se. But to go somewhere in the afterlife that we assume is happy. Perhaps back to the island, except this time it is a tropical paradise with monkey butlers, all the food they can eat, and fewer smoke monsters.

To be honest, I think it kind of worked. This may be because my expectations for the show dropped as I got through seasons 4 and 5, but I wasn't really disappointed. But what about the make-up of that group in purgatory? Like I said, there were definitely some conspicuous omissions. Eko probably didn't show up for practical reasons --- I suspect the actor wasn't interested. The actor playing Walt probably looks crazy old by now. But what about Michael? I guess he said his ghost was stuck on the island. I guess. But I didn't Lapidus either (full disclosure: I love Lapidus)? Miles gets to tag along, but nobody thinks about inviting Lapidus? He is the only reason any of you ever got off the island. He saved some of you TWICE! And then there is Sayid, who is just so happy to have found Shannon again. Do things just get really awkward when the two of them run into Nadia
in the afterlife?

Your turn, good sir.

Rick

I do love that movie

Then the Stallion sayeth 'If I can change, and you can change, then everybody can change.' And with that decree, the peoples of the world rejoiced, for the war, known for being of low temperature, had ended. Thanks to the Stallion and his eye, like that of a tiger.

     -- from Paulie's Letters to the Philadelphians

Saying It Better Than I Ever Could: Malcolm Gladwell on Writing

One of the things I have wanted to do with this blog is to highlight particular moments of clarity by writers, filmakers, artists, scientists, or really anybody who captures an idea/notion/belief that I've had/have, but could never express nearly that well. Today, I turn my attention to Malcolm Gladwell.

Today, I have demonstrated my incredible capacity for procrastination. In order to avoid my work, I have instead done the following: washed the dishes, knee stretches/exercises (which is usually something I procrastinate from and haven't done in months even if I am supposed to do them weekly if not more often), general web surfing, and hey, looky here, a blog post. Any way, in my continuing attempts to avoid "coding" (actually Mizar for those in the know), I picked up a book of essays by Malcolm Gladwell that, let's go with a somewhat obscure reference and call her Lula Mae (at least for now), lent to me. So far, I have only read the preface. In it, Mr. Gladwell drops the following piece of brilliance on us:

"The trick to finding ideas is to convince yourself that everyone and everything has a story to tell. I say trick but what I really mean is challenge, because it's a very hard thing to do. Our instinct as humans, after all, is to assume that most things are not interesting... We filter and rank and judge. We have to. There's just so much out there. But if you want to be a writer, you have to fight that instinct every day. Shampoo doesn't seem interesting? Well, dammit, it must be, and if it isn't, I have to believe that it will ultimately lead me to something that is."

--- Malcolm Gladwell. "What the Dog Saw and other Adventures." Little Brown and Company. 2009.

Gladwell made this remark in a discussion about where his ideas come from. I am obviously not a professional writer, but this is certainly something that I have found has plagued me when trying to come up with ideas for pieces for this blog. When it comes down to it, I don't believe that I am a particularly interesting person. And I don't necessarily mean that in a negative way (well ... maybe slightly). But simply put, I am not James Bond, nor a swash-buckling pirate, nor a jet-setting world-famous explorer. However, this does not mean I cannot write interesting things. Not that I do (or at least, that is up to you), but it is theoretically possible.

I have always aspired to be a better storyteller than I am. Since I am not a professional writer, the opportunities I have to research the lives of others and write about them is limited. But my ability to write about my own experiences is not. The challenge is to take my relatively straightforward existence, and make it interesting to my audience. Discussing the work that I do in a way that is both accessible, and, almost more importantly, exciting to those without the same background as me, is another challenge.

Sometimes I have succeeded in spite of this challenge, but more often I have not. Certainly practice helps in finding ways to make the seemingly unexciting into something interesting. Luckily, I have no shortage of things to procrastinate from.

Back on the Horse or The Continuing Misadventures of Rick

I've kind of neglected this space for awhile. But with this post, I am back with a vengeance! By which I actually mean I will probably continue to write things poorly, sporadically, and not very often.

It's been a very busy week, and in amidst all of my work, I forgot to pay my rent. This occurred for a few reasons. First of all, my memory is comparable to that of an ant. Secondly, February is a short month. Thirdly ... well, I don't have any more. And that second one is weak. When it comes right down to it, I'm just an idiot.

I eventually did remember, but it was not because of any inspired moment of responsibility. No, it was because I needed a 10 minute break from work, and so I picked up the ole guitar. I started to play a few tunes, and somehow picked out "Don't Worry, Be Happy" as a nice place to start. And when I got to the line that said "The landlord say the rent is late," I froze in recognition of my own stupidity.

All I know is that without Bobby McFerrin, my late fee probably would have been even higher than $25 I had to fork over today as I wondered exactly how it was that I function in normal society.

So ... Who Wants to Read About My Research?

I haven't been particularly active in this space lately, and I'll try to rectify that in the near future. In the meantime, I thought I'd post the introduction to my Master's thesis so that anybody who is curious as to my area of research can have a look.

As it is only the introduction it is mainly non-technical. I've also made some additional alterations in an effort to make it as accessible as possible for a non-computer science audience. Anyways, here goes:



Consider the problem of finding a route from one location in a city to another when navigating with a map. Before any travel can begin, a route for travel (or at least a partial route) must be found. Unfortunately, there may be a large number of candidate partial paths that must be considered before a complete route is found.

In this example, a single agent has the task of path finding. In general, all agents — whether they be living or artificial — are faced with some number of tasks to perform. In order to complete these tasks, agents must develop plans for action.
  
The effectiveness with which an agent plans is evaluated in terms of two different measures. The first is the time required to find a plan that will complete the given tasks. The second is the cost of the plan, the value of which will depend on the agent’s objectives. In the case of navigation, the possibilities for plan cost include distance travelled --- if the agent is to find a short path --- or the expected travel time --- if the agent is to find a quick path (ie. when navigating in a map, such an agent would prefer the use of highways over city streets).

In many domains, there is additional information that can be used to inform and thereby speed up planning. For example, when navigating between locations in Edmonton, Canada, it is reasonable to initially disregard routes through distant locations such as Madagascar. Moreover, if the desired destination is east of the starting location, the first routes to consider are naturally those that initially proceed eastward (if such paths exist).

Such information can be used to build heuristics which estimate the cost of the remaining path to the goal from any area in the domain. The field concerned with the development of heuristics and the construction of algorithms that use heuristics for planning is called single-agent search. In this field, planning is performed by building complete plans from some partial plans by a process which is guided by the given heuristic.

Single-agent search remains an important field for research due to the large number of real-world applications in which search algorithms have proven to be effective problem-solving techniques. These applications include autonomous robot navigation, in which heuristics are used to
guide pathfinding [6]; DNA sequence alignment [5]; and, computer games [7].

There is a class of single-agent search algorithms that will provably find the lowest cost solution provided that the heuristic satisfies certain conditions. Unfortunately, as problem domains grow larger, these optimal algorithms will often take too long to find a solution. To address this issue, suboptimal single-agent search algorithms have been developed. These algorithms sacrifice solution quality for a decrease in search time and are ideal when a solution (often near optimal) is needed quickly.

When constructing a single-agent search system, it must first be determined if suboptimal solutions will suffice or if optimal solutions are desired. Another consideration is if planning will precede execution or if the system is to work in real-time, with planning and plan execution being interleaved. The system designer is then faced with decisions regarding the proper selection of an algorithm and a heuristic function for the given domain(s). There are also often subtle choices such as tie-breaking (the order in which equally promising candidate paths are considered) that can significantly affect the planning speed. With all of these possible choices, properly recognizing and evaluating all of the necessary design decisions is a vital aspect of building an effective search system.

In the case of suboptimal problem solving, additional options arise as almost all applicable algorithms involve some kind of parameterization (where a parameter refers to some sort of numerical setting that can affect algorithm performance). For example, in the weighted variants of A*, IDA*[3], and RBFS [4], the value of the weight parameter must be set. Similarly, beam-search variants like BULB [2] and Beam-stack search [8] require the selection of a value for the beam width parameter.

Any adjustment of these parameter values can change both the solution quality and the search speed. In a few cases, there are theoretical results that indicate how changing a parameter will affect the search (such as the bounds on solution quality in weighted A*, weighted IDA*, and weighted RBFS [4]). Unfortunately, it is much more common that a significant amount of pre-computation is needed so as to determine how different parameter settings affect the search. In practice, parameter values are tested on a set of training examples through a process commonly referred to as parameter tuning. The single parameter setting that satisfies any given constraints on solution quality and exhibits the best average performance is then used in all future runs of the algorithm.

One drawback of parameter tuning is that it generally customizes the search algorithm for a specific problem domain (ie. an algorithm used for navigation may be tuned specifically for city-driving as opposed to for highway driving). As such, the results of this expensive process cannot effectively be transferred across domains — a fact that is of particular concern when designing general search systems such as automated heuristic search planners. In these systems, general heuristics are used to guide a suboptimal search algorithm over many domains. In practice, planners such as HSP [1] commit to a single parameter value that will hopefully be effective over a diverse class of problems.

Parameter tuning also suffers from another deficiency: there is no guarantee that a tuned value will perform well on each individual problem. Tuning only finds the setting that has the best average performance. On a per problem basis, there can be other parameter values which significantly outperform the tuned setting. In this thesis, we will demonstrate that it is often the case that if a search system could properly select the correct parameter setting for each problem, the planning speed over a number of problems can be greatly improved.

In this end, we will consider the dovetailing procedure as an approach to this problem. Dovetailing involves running several independent instances of an algorithm — each of which has a different set of parameter settings — at the same time by interleaving the execution of the instances. The
procedure can also be trivially extended so as to work with multi-processor computers. This aspect of the algorithm is important due to the increasing availability of such machines. As such, we will also analyze the performance of the algorithm when used in this fashion.

The main contributions of this thesis can be summarized as follows:

1. The single-agent search algorithms of WA*, WIDA*, WRBFS, and BULB are described, and the weaknesses of WA*, WIDA*, and WRBFS as suboptimal search algorithms are demonstrated. The behaviour of these algorithms in the sliding tile and pancake puzzles as a function of the weight parameter will also be shown. Similarly, the behaviour of BULB in these domains as a function of the beam width will be demonstrated.
2. Dovetailing is explored as an approach to proper parameter selection. Dovetailing is shown to significantly enhance WIDA* in the domains of the sliding tile puzzle and the pancake puzzle. The parallel version of dovetailing is also shown to exhibit massive improvements in search time when used with this algorithm. In the case of WRBFS, dovetailing without any pre-computation is shown to improve the speed of the algorithm in the sliding tile puzzle domain, and offer comparable performance in the pancake puzzle domain. Dovetailing with WRBFS over different tie-breaking schemes will also improve the quality of the solutions. We will also demonstrate that parallel dovetailing offers an effective parallelization of this algorithm.
3. The sequential version of dovetailing will be shown to decrease the search speed of BULB and WA*, but improve the quality of the solutions when considered over different tie-breaking schemes. The reasons for this behaviour are discussed. The parallel version of dovetailing will also be shown to offer the best performance of all known parallelizations of WA* when higher weights are used in the sliding tile puzzle domain.

Works Cited

[1] Blai Bonet and Hector Geffner. Planning as heuristic search. Artif. Intell., 129(1-2):5–33, 2001.
[2] David Furcy and Sven Koenig. Limited Discrepancy Beam Search. In IJCAI, pages 125–131, 2005.
[3] Richard E. Korf. Iterative-Deepening-A*: An Optimal Admissible Tree Search. In IJCAI, pages 1034–1036, 1985.
[4] Richard E. Korf. Linear-Space Best-First Search. Artif. Intell., 62(1):41–78, 1993.
[5] Matthew McNaughton, Paul Lu, Jonathan Schaeffer, and Duane Szafron. Memory-Efficient A* Heuristics for Multiple Sequence Alignment. In AAAI/IAAI, pages 737–743, 2002.
[6] Anthony Stentz. The Focussed D* Algorithm for Real-Time Replanning. In IJCAI, pages 1652–1659, 1995.
[7] Bryan Stout. Smart moves: Intelligent Pathfinding. Game Developer Magazine, October:28–35, 1995.
[8] Rong Zhou and Eric A. Hansen. Beam-Stack Search: Integrating Backtracking with Beam Search. In Susanne Biundo, Karen L. Myers, and Kanna Rajan, editors, ICAPS, pages 90–98. AAAI, 2005.

By the way, if you'd like to read the whole thing, send me an email and I can pass it along.

ALERT: Blue Jays Rant Coming!

Today's been something of an annoying day. I had to wake up at a reasonable hour as if I have a real job. Then I found out that my move to another apartment is being bumped back to a much more inconvenient date. And all day I have been working on an assignment and it ain't going well.

Long story short, I am in a ranting move. So here goes ...

A few days ago, Keith Law mentioned that when he was with the Blue Jays, JP Ricciardi turned down a trade of Jose Cruz Jr. (retired centre-fielder) for David Wright (current perennial all-star and MVP candidate). Now, in hindsight this was an abysmal decision, but it was at least kind of defensible at the time when Jose Cruz was viewed as being a strong defensive player who might develop his plethora of tools.

I'm not here to apologize or pile on JP; I'll leave that to others. But I do want to know why Keith Law had to confirm this rumour. Does he just hate Jays fans? I mean, the rational side of me (small as it may be) is trying to argue that these non-trades happen all the time, and prospects are always a bit of a gamble.

Unfortunately, the sports fan in me wants to go on table-flipping rampage (which has been shown by science to be the second best kind of rampage after the Rambo-like rampage). DAVID WRIGHT!? REALLY?! Keith Law, why must you taunt me? It's bad enough that the team this year was terrible, but you just had to poor the salt that is David Wright on those wounds? Next, you'll remind me about the Lincecum for Rios non-trade and that Roy Halladay is probably leaving. Or that the Leafs are 1-7-1, Toronto FC just got embarassed in a do-or-die game, the Argos are a disaster, and the Raptors may go games without getting a single rebound. I really don't need this right now, Keith Law, I really don't.

I also don't want to kick Steve Phillips while he is down, but this is just more evidence that he was a terrible GM. At least JP can say he wasn't familiar with Wright --- Phillips doesn't have that luxury. And yet, when he goes on television as a baseball analyst, I'm supposed to take his opinion seriously?

But let this be a lesson for new Blue Jays GM Alex Anthopoulos. Alex, do yourself a favour. Determine who are the worst GM's in the game (I suggest starting with that fella in San Francisco), and get on their good side. Then call them weekly with ridiculous trade offers. Particularly, offer aging players whose skills are degrading for shiny young prospects. Somehow these trades happen far more often than they should, and its about time the Blue Jays benefitted from the stupidity of other teams.