social media analytics

In this post we study the growing field of social media analytics. In order to address this at least somewhat satisfactorily, we will consider the outline below. All topics will be covered somewhat superficially for the cognoscenti, though in enough detail for the not too deeply initiated, so keep in mind this might either be rather basic, or quite sophisticated, depending on where you are coming from, if you decide to read it.

What is Social Media (for purpose of this post)?

By social media here we refer to all media that exhibit three characteristics:
1. low barrier to entry technology
2. widespread adoption
3. preferably a bi-directional communication channel

What do we think of social media given our definition: the blogosphere, yelp, Twitter, Facebook, Google+, recommender systems like Amazon and Netflix, etc., What are the technologies or media we speak of here? Mostly the Internet (WWW) and Mobile.

What do we mean by Analytics here?

We define Analytics as a means to analyze data to extract intelligent, actionable inferences that will enable us to generate "an edge". You can build an edge in any environment if you collect, analyze, and interpret the correct data correctly. The challenges here are many-fold - finding the right data to collect, then collecting the right data, then identifying the right set of analyses to run on that data, running the analyses correctly, generating intelligent, usable, actionable inferences therefrom, and finally applying the correct actions to build, develop, and maintain an edge in a business to profit from, and leave your competitors behind. Whew! if that sounds like a tall order, it is. Analytics is hard to get right, but also extremely important.

Social Media evolution

Social Media, by its very definition, is a networked platform. This means to grow a social platform, one needs more users over time to offer them services, and one needs services to attract and retain more users - so somewhat of a catch-22. Some of the most successful businesses are Network Platforms. 

• Google revenues increase many-fold because it is a search platform that also offers advertising. The more it is used for search, the more eye-balls that will see the paid links that appear on the top of search results pages. The more there are ads, the more attractive searches may appear to be especially to people that are looking to buy particular products.
• Amazon's sales muscle comes from hosting a fantastic platform where it can act as a digital shopfront at the same time gathering actual user reviews and feedback - this is useful not just to potential buyers, but also engenders a feeling of community. And this community attracts more sellers. Recently even Ikea started selling its wares over Amazon. It's a win-win.
• Facebook brings users together. These users first came together to build a community in a simple platform with limited services. But as users grew, the medium became an interesting place to host/add new services. Now FB is an ecosystem, a platform on which new apps are built - and many other companies, like Zynga, Playdom etc use it as a substrate to grow their own businesses.
• Twitter - the most prominent micro-blogging service today, helps people share meaningful short commentary, with the world, or a select group of followers, as they like. This works particularly well given how the new digital generations multi-task more, have shorter attention spans, and quicker context switches. We don't tend to think of the old SMS system in telephony as Social Media, but it is based on the same principle - SMS even had VASPs or Value Added Service Providers that were able to deliver jokes, horoscopes or other media of interest to subscribers... for a fee.
• The number of social media platforms is increasing exponentially (WhatsApp, weChat, Telegram, ...), which is funny when one considers that the number of connected people (these have to have access to technology, sometimes leading edge technology, and large parts of the world do not have this yet), while growing, isn't growing as rapidly. Which means social media penetration among the population that has it is greater than 100% (this remark is made only somewhat tongue-in-cheek).

Uses and Applications of Social Media [Examples to be provided later...]

• Education - social media debates, multi-player educational games, tutorials over Skype etc
• Advertising - targeted location based advertising, opt-in ads, micro-targeted ads etc.,
• Marketing - word-of-mouth marketing, trailers going viral, recommender systems etc
• Sales - recommender systems, price discovery etc.,
• Entertainment - finding suitable content socially, viewing together from remote locations etc
• Services - e.g. weather forecasts, severe weather warnings, reverse 911 etc
• News - news reporting, flash-mobs, etc 
• Finance - investment analysis, social media sentiment analysis etc

Risks with Social Media Technologies
Typically, Internet Services need to have at least a simple security architecture in place. This means having components that account for functions such as authentication, authorization and accounting (collectively known as AAA) and non-repudiation, built into the system.

• Authentication is a process that ensures that you are who you say you are when you request the service. A simple example of this is the use of user-name and password.
• Authorization is the process of determining that you the authenticated user, has the right to perform the operations you request. For example, if you request that a particular computer log file be deleted, the computer system might not let you perform this action unless you were the systems administrator.
• Accounting is the process of tracking the authorized actions you perform on the system, and then being able to charge you for these later, within the defined parameters of a billing model or system in place.
• Non-repudiation is an add-on to authentication in that it ensures that once you say or do something, you cannot later disclaim having done so. In other words, your action is provably tied to your authenticated identity. An example of this is, if Bob digitally signs a document with his private key and sends it to Alice, if Alice holds him to the contents of that document at a later date, Bob cannot say he did not sign that document since only he is supposed to have knowledge of his private key.

A major issue with social media today is that not all of these components are available in all systems, even those with millions or hundreds of millions of users. Users casually and sometimes callously share their most personal and intimate details online, and these can be exploited by unscrupulous elements that abound on the Internet, either for personal gain or for defaming their character. This is a major risk that is quite important in the current social media application setting.

How to mine data from Social Networks
We cover this in a separate post.

The "feature interaction" problem - potential Chrome and Android (and iPhone) issue

In traditional telephony, the Feature Interaction Problem (FIP) is a major concern when code for switches is developed, particularly as more features are added to particular subscriber lines. For example, Bob might have call forwarding from his home phone line to his office during work hours, but has Do Not Disturb enabled on his office phone line when a call arrives on his home line. This results in indeterminate treatment for the incoming call. In order to avoid these kinds of situations, wherever possible, phone companies used to implement careful priority rules for how particular features might work.

A more complete explanation is here: http://www2.research.att.com/~pamela/faq.html

While this problem might be mitigated somewhat in the context of a single phone line, it gets much more involved and difficult when multiple users or phone lines are involved (e.g. in the scenario from the previous paragraph). For instance, if Alice has both Call Waiting and Voice Mail features enabled on her line, a simple priority setting may be applied giving Alice up to 4 rings to pick up the waiting call, after which it is forwarded on to Voice Mail. Features in telephony switches are typically linked to triggers in the Call Model, so these can be set to work in a pre-defined priority order if there is risk of interaction.

There are several papers on this topic particularly from people who worked on the 5ESS and other switches from AT&T Bell Laboratories, long the premier Research laboratory in the world.

So why is the FIP an issue today? Let's see. In the context of Android applications or Chrome browser extensions, there are several unrelated applications a user is permitted to install and activate. Several of these might request and obtain user permission for various features including call treatment etc., unless the user is very careful, it is quite possible that indeterminate, or unexpected behavior might occur that might at best, reduce the user's delight with the use of the system, or at worst, make the system unbearably painful to use. Admittedly, this is less of a risk in the initial phases when the number of applications that utilize the more involved features supported by the user's device are available, but as these proliferate, the risks tend to rise.

Note one other thing here - in the older scenarios from traditional telephony, the location of the problem as well as the capacity to apply a fix here, were both centralized at the switch so the phone company could do these for you. This is no longer the case on the advanced smart phones of today. The issue is now localized to the smart end-point, and becomes something the end-user has to fix for themselves. And end-users tend to be far less savvy about these things than phone company Research staff tend to be. Besides, different end users have different applications installed, which might result in different feature interaction issues being manifest.

One way of minimizing these kinds of issues might be to design the APIs that developers can use such that these kinds of interactions are minimized by limiting access to particular application triggers (e.g. when apps are sleeping and waiting on a particular event to happen), or by forcing the end-user to define a priority order among these various applications on the phone. This would require a Feature Interaction Manager or FIM application to ship with the phone or as part of the smart phone operating system so that undesirable user scenarios do not occur, or their occurrence can be strictly limited to a small number of cases.

So is this all just an academic exercise at this point? Not in the least. There are several instances where adding too many extensions and enabling them all simultaneously in Chrome causes the browser to only display blank pages. No matter what URL you type, or what you do, you only see blank pages. This is debugged and fixed by turning off and then back on, extensions one at a time. This fixes the issue and makes the browser usable again. Lots of people report this issue (examples below).

Sometimes this is a result of rootkits or viruses: e.g. http://www.wintips.org/fix-google-chrome-blank-page-problem/, but other times it is resolved by turning off and on extensions as described above.
http://forums.androidcentral.com/google-nexus-7-tablet-2012/263547-blank-page-chrome.html

Since there is no easy way to fix this problem in the most general way (short of perhaps a FIM), perhaps more careful API design, developer education, and user population education that this problem exists, and how it might be addressed when it occurs, is a good way to proceed.

Google Drive - problem and solution (advanced use case)

Many users have a very common complaint with using Google Drive, which goes along these lines:

I signed up for Google Drive and when I want to sync a folder that already contains files, the program complains that the directory is not empty. But I already have two hundred gigabytes of data in my C:\ABC directory and programs that use that data with that directory hard-coded in. Now if I move all my files over into the Google Drive folder as the Google program wants, I have to re-write all my directory paths in my code again. And if I don't move my programs to the Google Drive folder, Google's drive program won't sync with the cloud. Why can't Google fix this?

See for instance: http://productforums.google.com/forum/#!topic/drive/-TmjPhFSNAY

Thing is, Google doesn't need to fix this. Of course, it would be nice if Google's Drive sync program were able to sync a non-empty folder into the cloud, especially for new accounts. But you as a user are not helpless. A fairly simple remedy exists. This works for both Unix variants and Windows, but this post will cover how to solve this issue in the Windows environment. USE AT YOUR OWN RISK. This worked well for me.

The key to solving this problem is an idea called a hard-link. The way this works, if you create a hard-link from one directory A to another newly created directory B, then this creates a virtual copy of the directory A in B. In other words, C:\A and the newly created C:\B point to the same contents (namely the files and folders under C:\A), and any updates made to C:\B\abc will be reflected in C:\A and vice-versa. Hard links are also sometimes called "junctions" in Windows parlance.

Now, how to solve the problem in italics from above? Until Google fixes their Drive program to be able to sync any non-empty folder(s) into the cloud, what you can do is:

1. backup the data you want saved into the Google Drive (C:\ABC) to an external hard drive device
2. install Google Drive, set a location for the Google Drive folder or stick with the default proposed location
3. move (do not copy) C:\ABC to Google Drive folder from (2) above
4. the Google Drive program is happy with this, and starts sync'ing the files and folders to the cloud
5. wait for the move in step 3 to complete. verify that it was completely successful. then delete (remove directory) C:\ABC.
6. create a hard link between the Google Drive folder from (2) above, and C:\ABC. this step will only work if the C:\ABC drive was completely removed in step 5. 
7. Google Drive will now sync all your files, and your programs that were hard coded with the C:\ABC path will continue to work.

So now we need to discuss how one might create a hard-link to a directory in Step 6 above. This is easily done as follows: 

1. Open a command prompt as administrator (right click on your command prompt icon in the Start Menu, then pick "run as administrator"). 
2. type in "mklink/?" in the command window and hit enter. this will show you the various options or switches you can use with this command.
3. to create a hard link to a directory, you need to use the following syntax: "mklink <link> <target>". if there are spaces in either of your directory names, you need to use quotes around it (as usual). So for instance, here you are creating a hard-link between your target directory of "C:\Google Drive" (or whatever drive name you picked for the Google Drive, and C:\ABC (which now doesn't exist on disk anymore). So you type: mklink C:\ABC "C:\Google Drive" and hit <enter>. This creates a "junction" or hard-link.

... and you are done.

So how do you delete a hard-link? Deleting a directory hard-link is the same as deleting a directory in Windows. Simply type "rmdir C:\ABC" at a command prompt, and hit enter. Remember, you specify the link directory, not the Google Drive directory, in the command above. This way, your data still exists, just the link is removed.

Interesting links for comparison of various cloud drive services:
http://macography.net/2013/05/speed-test-dropbox-google-drive-box-skydrive-amazon-cloud-drive/
http://www.theverge.com/2012/4/24/2954960/google-drive-dropbox-skydrive-sugarsync-cloud-storage-competition

The Many Headed Hydra - Sci-Fi short story

This is from my novella, a work in progress...

"A soul is a reflection of Infinity's light in a finite form." -- Ancient Dharmin text

The creature with no name, one we will call Gestalt, moved across the galaxy. He was as old as time, with accumulated experience (wisdom?) and a desire to live peacefully on his own. After many aeons on a planet, he saw life develop on other rocks in the system, and decided it was time to move - space was too vast to have to live in crowded confines.

Gestalt loved hospitable zones - ooh to bask in the light of a star or two or three. And avoided black-holes like the proverbial plague, though the plague itself could do nothing to him. A narrow escape millenia ago where he lost nearly a sixth of his mass to one, left him wiser to the ways of the universe.

He hated moving "house". As first tenant in many ways, he could not comprehend why he ended up shifting when new life sprouted in his neighborhood. For a while things were usually tolerable and other life-forms left him alone. But every once in a while they would intrude on his space... and shy as he was, he would usually just take off through space for another living zone. "My space within space", he thought. And might have laughed too, but we do not know if he was capable of levity.

What is time really on a cosmic scale with no clock to measure it by? Sure, intra-atomic vibrations kept time, but "time" at that scale meant nothing to Gestalt who was as old as the idea of infinity. After traveling for "a while", he (it?) arrived at a rock on a system he deemed comfortable. Slowly rotating, with two stars for warmth, and only two other planets somewhat nearby, Gestalt felt sated and content. This was a nice corner in universe with no corners - this would be his home forever now. And so he lived... eking out sustenance from the warmth of the suns.

Gestalt was starting to feel lonely. As far as he knew, he was the only one of his kind. He knew nothing of his origins or of his purpose. But if he was life, and there was other life, maybe he could talk to it? That thought would now have warmed his heart - if he had one.

Millenia passed, and once again, he perceived life sprouting on the other two planets. This time, rather than just leave, Gestalt decided to adapt. What had he to lose? Other than the black-holes, he had not seen any entity that could hurt him in any way. A simple experiment couldn't hurt now, could it? And besides, life usually took a while to move off any one rock to get to a point where it could reach him. So time was in his favor.

Indivisible though he was, he simulated a liquid with his structure, and evolved it into a fixed number of separate anthropomorphic life-forms the likes of which he had seen on many planets before. Each of these was imbued with its own "personality", but they were all part of him really. They retained unique identities but a voluntary collective consciousness. They could share experiences with each other (and in so doing, with him), or not - as they chose. But connecting to the collective assured them sustenance even in this evolved form. Gestalt was particularly proud of his engineering free choice into the mix.

Since there was only so much of him to go around - in terms of physical material - there were a fixed number of forms on the rock. They would associate with each other at will, but if a form tired, decayed or felt hurt, he would absorb it into himself, enriching the collective consciousness with its private experiences, and "regrow" it through the stages of development he had witnessed everywhere there was life. Every freshly issued creature was born with a "veil of nescience" -- the knowledge of the collective was only gradually revealed to it -- though birth came with new vigor and energy to once again embrace and enjoy its existence to the fullest.

Gestalt let his "children" evolve - they, who used to communicate almost exclusively telepathically at first, created a language, then clothing, crude implements, then fire, more sophisticated tools, a social fabric, better and more comfortable dwellings, means of moving around the planet, ... yes, some aspects of this felt like devolution - using talk as a crude replacement for telepathy, for instance - but he convinced himself this let him more aptly mimic other life-forms he had seen before. To embrace his new reality fully, he gave it a name - Dharma... his children called themselves the Dharmins.

If one Dharmin "ate" (absorbed light and warmth for sustenance), satisfaction spread through the collective proportionately. If one rested, the collective felt refreshed. Gestalt felt he did well by them - gave them the means to eliminate want, while giving them (himself in their form) the ability to enjoy individual experiences with a modicum of privacy and detachment... and so life went on, until Dharmins became from the one, many.

They still had a collective consciousness, and telepathy, but gradually became more individual, more specialized - because they could adapt differently. Young ones were raised by the community. The Elders cared for them, guiding them in lifting the veil that for a while obscured their complete glory, while the others worked to improve society. With Gestalt's wisdom and the knowledge of the ages, they knew every nook and cranny on their rock and lived in harmony with their environment. A peaceful, genteel people ignorant of
want or war, their life was a continual renaissance.

A young Dharmin to another: "Does Gestalt really exist?"
The second answers: "Gestalt only knows."

The Dharmins mined the rock for a mineral called Kesavanium found plentiful on that planet, and built a shrine to The One Who Might Not Exist. They would call it Gestalt's Temple. Gestalt knew he had done well when he saw this monument to him, his "children" had constructed. A large three dimensional holographic inscription on a wall in the main hall of this crystal monument read: "G E S T A L T I S N O W H E R E". Make of it what you will.

A venerable, ageless form called The Sage - a true seer if there ever was one - presided over this monument, greeting fellow Dharmins that visited. He was always fully immersed in the Gestalt's being and spoke in riddles if he spoke at all. Many legends were written and ballads sung about him - legends that transcended the history on that rock.

A comparative analysis of Recommender Systems - Amazon and Netflix

Recommendation systems tend to add significant value to a web-store, particularly since a. they can increase traffic towards less demanded items in the store, and b. they can enable you to leverage your existing customer base to "market" to users considering purchases of new goods related to ones other customers with similar tastes have bought in the recent past.

There are however several non-technical aspects one needs to be aware of as one designs recommender or recommendation systems, among them the following:

1. recommendations provided must use a very "light touch" and indeed be relevant, or users will bail (recall that users have very low switching costs in the Internet/web context)
2. it is usually a good idea to tell users on what basis a recommendation is made, to reduce annoyance or increase delight at the use of your systems - e.g. are we recommending a product to you because others like you (similarity defined in any reasonable way - similar profile etc) liked that product? because others who bought the same products you did, bought this other product? because others with the same browsing history as you bought the product in question?... and other similar considerations
3. ease of rating and generating reviews
4. ease of user-base interaction amongst themselves (the more vibrant your user community, the less hand-holding you might need to do, if you have good automated systems to moderate user forums etc)

Also, a key difference here is whether you are selling your own products, or acting as a store-front to others' products - in the latter case, you can of course be much more open and neutral in terms of the user-generated content that is hosted. And quite frankly, I don't think I have seen a good example of the former kind of systems in practice.

That said, on to the topic at hand... Both Amazon Prime Instant Video (APIV) and Netflix Video (NV) have recommender systems built in, but they differ in some important ways:

1. APIV says "people who watched what you just watched" also watched these titles A, B, C... While NV says "people who liked the title you just watched, also liked these titles E, F, G." 
2. APIV as perhaps the largest and most widely used store-front today (sorry EBay) knows a lot about what you've bought, and can leverage that to make recommendations. Of course, APIV doesn't know whether you liked what you bought or watched already in all cases. NV only knows what movies you have seen and which ones you liked (based on what you've rated or said you liked) and can use that as a basis for making recommendations.
3. Both leverage information about user behavior. Neither uses information about the product. For instance, new movies, before anyone has watched them, will have zero stars in both NV and APIV. 

Between the two, you could argue that NV offers the more targeted recommendations primarily because these are tuned to your taste, based on what you've told NV already that you liked. APIV's recs, on the other hand, are much more generic, and are more focused on getting people to watch videos using their free subscription, and arguably less focused on whether this is what people might want to see.

NV has the better store-front for renting/streaming video - the popups on mouse hover over movie icons is something users expect, and it is disappointing to see Amazon doesn't implement yet. But then Amazon has hands-down the better infrastructure at the back-end - don't forget, even Netflix is hosted off Amazon servers. But it is user experience we are talking about here, and NV wins.

It is not hard to envision a world where a recommender system is stripped off a store-front and is run by a third party. This puts various streaming store-fronts on a level playing field where the main focus here is efficient user choice, and potentially, cost. We have less time to waste these days given our busy schedules, and have to choose from a larger variety of products more quickly. A system that helps us make the correct choice more efficiently will drive storefront success. A system decoupled from a store-front owner gives a sense of neutrality and unbiased-ness which is critical to its user acceptance.

Billing becomes more complicated in the latter scenarios however. Currently the user pays the service provider for the service. In the latter case, the user may pay the service provider who kicks back some amount to the middle-man for referring users to their site, or the user has to pay a fee to the middle-man for access to better sites. The former model has similar issues with neutrality and conflict of interest as one saw with Ratings Agencies during the recent Financial Crisis and debacle. The latter... well, it complicates matters significantly. And no one wants to pay more for something they're accustomed to getting for free now. Look at the challenges New York Times and now Washington Post are having at monetizing their digital content.

It is a pity that neither NV nor APIV utilize recommender systems that leverage thematic elements in movies. For instance, one can compute statistically improbable phrases or SIPs within either movie synopses or in movie content (e.g. subtitles) and try to determine other movies with similar content that watchers might like. TF-IDF based systems that do this might not be too difficult to build.

So we didn't mention Hulu in the above. Well, that business model is slightly difficult to comprehend. The most common gripe I hear from even fans and subscribers of the website is that even after paying for premium membership, the user is still subjected to ads. True, the "premium" aspect provides for easier access to full seasons of shows, but people feel cheated that after paying for an "Internet" experience, they are still treated like "ordinary" TV watchers. After all, if you allow time-shifting and place-shifting, you should permit the more discerning users to watch videos the way they want as well, especially if you're charging them.

A day in the life of an FX and Rates Strategist

Strategists in the financial services sector are of different types depending on the asset class they cover and whether they operate on the buy- or sell- sides of the business. This role is different from the one business strategists perform in various firms in that it is focused on capital markets.

Typically a strategist does not trade, but comes up with (strategizes) trade ideas that the traders can put on and portfolio managers can consider. Usually strategists look at various types of indicators as they perform their duties - indicators like macroecnomic news and numbers releases (e.g. India's RBI likely to intervene in FX markets by selling $10 B to help bolster the rupee), quantitative signals (e.g. the volatility of MYRUSD returns is much higher than that of the USDIDR), and technical indicators (e.g. the ichimoku cloud indicates that the move in USDJPY is well supported with a likely move yet higher in the coming days).

As with all else, strategists also vary widely in quality. Here, we focus on the role of FX & Rates strategists on the buy-side. A typical day might be something like this:

0700: Get to work, read up news stories on what happened overnight, and potential changes
          to views currently held. Also, familiarize yourself with the current trade book
          including trades executed in the overnight hours

0800: Look across markets to determine where any important economic numbers are going to
          be announced. Prepare notes for the morning call.

0830: Go to morning call, discuss current positions, corrections to previously expressed
          views, pending news and numbers releases, changes in technicals for various markets,
          favorite trades if any etc

0900: Study Bloomberg screens and sales side analysts' research to identify and analyze
          potential trade ideas.

1100: Write-up investment theses carefully focusing on which trade ideas make sense for
          which funds (since different funds are run to different mandates), and in what
          market contexts. Focus on what the entry and exit levels are appropriate.

1400: if you manage a paper portfolio, put your trades in there. track P&L carefully

1500: take calls from portfolio managers, pitch ideas to them, meet with sell-side analysts
          learn more about your markets, pending central bank actions, macro-economic
          developments etc

1800: send out end of day commentary, list important news stories and events to watch for
          the next day, and over the rest of the week, with updates.

A day in the life of ... A Credit Analyst

Credit analysts track various sovereign, quasi-sovereign or corporate credits (i.e. companies). They model the names they track along various accounting metrics to determine how the companies are currently performing, and how they might perform under different market conditions. They advise portfolio managers on what credits to buy and which ones to sell and when. Good credit analysts are able to anticipate market moves before they happen and steer portfolios to tidy gains while avoiding losses. Credit analysts tend to be very busy during equity earnings season (yes, even though they work in the corporate bonds space), and with new or potential new issues.

This is what a typical Credit Analyst's day might look like:

0700: get to work, track major market news stories and overnight 
      developments

0800: participate in the morning meeting with all other analysts 
      and portfolio managers present recent happenings in your 
      sectors, names you cover, market news items of interest. 
      discuss any relevant holdings in portfolios, summarize key 
      ideas for either increasing or decreasing exposure in 
      particular companies

0900: if on the sell side confer with other analysts that cover 
till  the same name or attend scheduled meetings or conference 
1100  calls with management from companies you cover. ask 
      pertinent questions to firm up your investment theses

      if on the buy side, talk to sell side analysts that cover 
      the same name or credit, ask for trade ideas either in 
      terms of relative value or others, and form firm 
      recommendations

1100: read research on companies, track market events to 
till  evaluate trade ideas factoring in portfolio holdings. keep 
1500  aware of new issuances from either companies you hold or 
      in sectors you cover. pay special attention to M&A     
      activity as well.

1500: pitch your trade and holdings ideas to portfolio managers. 
till  clarify views on your credits etc. it usually helps to put 
1600  out a note on names you cover daily before leaving work 
      with major news items and impending news stories. for new 
      issues, might want to advise portfolio manager on 
      allocations to put in for.

1600: build models for the companies you cover, test them with 
till  various scenarios if working on regional credits, might 
1800  also want to model various fat-tail events to determine 
      how credits might behave

1800: work on daily commentary, send it out, head home
till
1830