{{comic
| number = 1876
| date = August 14, 2017
| title = Eclipse Searches
| image = eclipse_searches.png
| titletext = There were traffic jams for the eclipses in 1970 and 1979, and that was *before* we had the potential for overnight viral social media frenzies.
}}

==Explanation==
This comic is the first of five consecutive comics published in the week before and during the {{w|solar eclipse}} occurring on Monday, {{w|Solar eclipse of August 21, 2017|August 21, 2017}} which was a total solar eclipse visible within a band across the {{w|contiguous United States}} from west to east, and visible as a partial eclipse across the entire contiguous United States. The other comics are [[1877: Eclipse Science]], [[1878: Earth Orbital Diagram]] and [[1879: Eclipse Birds]].

[[Cueball]] comes to tell [[Megan]] about a cool map showing that searches on Google on the word Eclipse trace the same path across the USA as the totality band does, implying that those living in the zone are more interested than the rest of the US population.

The "cool map" is hosted by ''{{w|The Washington Post}}'' and sourced from {{w|Google Trends}} data. The link shown in the comic is here: [http://wapo.st/2vkgIBv wapo.st/2vkgIBv] (subscription required); an archived version is available [https://web.archive.org/web/20170814171105/https://www.washingtonpost.com/news/wonk/wp/2017/08/01/the-path-of-the-solar-eclipse-is-already-altering-real-world-behavior/ here] at archive.org

Since the eclipse searches are outpacing the 2016 election searches now, this is saying the eclipse popularity is going to rocket upwards just before the eclipse. Cueball is thus warned by Megan that the extreme amount of social media interest in the eclipse may lead to massive traffic jams, as last days frenzy regarding the eclipse will cause an enormous amount of people to decide to go to the eclipse in the last moment, causing huge traffic jams. Also as soon as people driving on the freeway enters the totality zone it has been seen happening that people just stop their cars and get out blocking the roads. This time also the eclipse-viewers will wish to post their content on the social media which might also cause a cyber traffic jam, in which users may find that they experience delays in sending or receiving data due to a high demand on telecommunications infrastructure. Megan tells Cueball to bring water if he is on the road during the totality, the implication being that people who are on the road may be stuck in their vehicles for long periods of time, and thus need refreshments.

In the graph charting interest in the {{w|United States presidential election, 2016|2016 US presidential election}}, November 8 is an important date as it was the day the election was held. August 21, 2017 refers to the date of the then upcoming solar eclipse.

The title text refers to the total eclipses from {{w|Solar eclipse of March 7, 1970|1970}} and {{w|Solar eclipse of February 26, 1979|1979}} which were also visible in the US, but both only for a few states. The traffic jams will be worse than those caused by previous eclipses, as we did not have viral social media in the 1970s, and also much less traffic on the roads.

==Transcript==
:[Cueball walks up to Megan while holding his smartphone up in his hand. Megan sits in an office chair in front of her desk with an open laptop on it. She turns her head towards him. A footnote to Cueball's speech is at the bottom right of the panel.]
:Cueball: Have you seen this cool map* showing how Google searches for "eclipse" trace the eclipse's path?
:Megan: Yeah!
:Megan: But you know what else I noticed?
:Cueball: What?
:Footnote: *wapo.st/2vkgIBv

:[Two panels are on top of each other. In the smaller top panel a colored graph is shown with a blue growing graph and a steeper growing red graph above it. Both are labeled. The x-axis (a black line) has labels and arrows below it in red and blue to indicate time periods. Text above the graph:]
:Searches for "eclipse" in the weeks before the eclipse are outpacing searches for "election" in the weeks before the 2016 election.
:Red line: Eclipse
:Blue line: Election
:Red x-axis labels: 2017 →|← July →|← August → 
:Blue x-axis labels: 2016 - September →|← October →|← 

:[Behind the top panel is a second larger panel slightly lower end more to the right. The visible part at the bottom of this panel is showing the same line graphs with the one from the election going past the election day. The x-axis labels ranges over a later time. The blue graph has a huge peak at the election day, visible in the part of the panel to the right of the top panel) and this date is written in a blue dot on the label. Similar the date of the Eclipse is written in a red dot. The red graph above the blue still ends in dots before the expected peak, as it is in the future.]
:Red line: Eclipse
:Blue line: Election
:Red x-axis labels: 2017 - July →|← August 21 → 
:Blue x-axis labels: 2016 - October →|← November 8 → 

:[Megan with her hands on the laptop and Cueball are both looking at the laptop. Beat panel.]

:[Cueball stands again and Megan has taken her hands down from the laptop.]
:Cueball: This is gonna be bad, isn't it?
:Megan: If you're planning to be on the road next Monday, bring water.

==Trivia==
*This is the second reference to the Eclipse within a month, the first was in [[1868: Eclipse Flights]]. The 2017 eclipse was mentioned as early as 2013 in the title text of [[1302: Year in Review]]. And this years ''New Year'' comic [[1779: 2017]] also mentions it. Both comics joking in the title text that it may be canceled or not happening.

*It was the first time in about three months [[Randall]] made a reference about the presidential election from 2016.

{{comic discussion}}

[[Category:Comics with color]]
[[Category:Comics featuring Cueball]]
[[Category:Comics featuring Megan]]
[[Category:Line graphs]]
[[Category:Astronomy]]
[[Category:Total Solar Eclipse 2017]]

1881: Drone Training

2017-08-25T16:44:02Z

OriginalName: /* Explanation */

{{comic
| number = 1881
| date = August 25, 2017
| title = Drone Training
| image = drone_training.png
| titletext = The joke will be on him in a few weeks when animal control shows up and takes custody of his Roomba.
}}

==Explanation==
{{incomplete|A simple comic, anything missing?}}
In this comic, [[Black Hat]] enters a pet store run by [[White Hat]]. He wants to buy something to help him train his drone, which keeps flying into the wrong rooms. This is absurd as drones are not living creatures like dogs or cats, which can be trained to do tricks, or stay in the correct areas (inside his property). He also wants a shock collar for his {{w|Roomba}}, which would train it to stay inside or at least on his lawn.

{{w|Cat repeller}} are devices or substances for training cats or repelling them from furniture or other areas. The Wikipedia page lacks the spray but you easily can create it at home as shown here: [http://www.instructables.com/id/Friendly-but-Effective-Cat-Repellent/ Friendly (but Effective) Cat Repellent], Black Hat is probably wasting some money when he buys it in a shop.

Roomba is a branch of autonomous robotic vacuum cleaners and controlling it by electric shocks from a {{w|Shock collar|shock collar}}, normally used for dogs, is more than questionable. Those collars are legal in the US but they are banned in many European countries and others.

Knowing Black Hat, he might be poking fun at either people's assumptions that modern robots are more advanced than they actually are, or possibly at White Hat's various inaccurate claims. It is also possible, however, that he just wants to terrify people with the idea of robot abuse/conditioning. There is also a chance that he genuinely believes he can force robots to obey him via inflicting suffering, and would prefer doing so to just using a remote control on the drone, or an ordinary vacuum rather than a Roomba.

The title text predicts that it will become a total flop when the Roomba probably gets rid of the collar and terrorizes the neighborhood. As a result dogcatchers from the {{w|Animal control service|animal control service}} will arrest this ''wild animal''.

==Transcript==
:[Black Hat stands in front of a store counter facing White Hat who stands behind.]
:Black Hat: My drone keeps flying into the wrong rooms. Do you have anything to discourage it?
:White Hat: Sir, this is a pet store.

:Black Hat: Yeah, I was thinking one of those spray bottles for cats.
:White Hat: I don't think you can train a drone.

:[White Hat hands a spray bottle over the counter.]
:Black Hat: Eh, they say that about cats, too. Plus, these days they probably all come with deep learning or whatever. Drones, I mean. Maybe cats too.
:White Hat: Fine, here's a bottle.

:Black Hat: Do you sell a shock collar that can fit around a Roomba?
:White Hat: I'm going to have to ask you to leave.

==Trivia==
*This comic breaks a 5-comic, 9-day (August 14 - 23) eclipse streak, which included [[1876: Eclipse Searches]], [[1877: Eclipse Science]], [[1878: Earth Orbital Diagram]], [[1879: Eclipse Birds]], and [[1880: Eclipse Review]]. This is the longest of [[Randall]]'s same-topic streaks, as his other 5-comic streaks took place in one week with special Tuesday and Thursday releases.

{{comic discussion}}

[[Category:Comics featuring Black Hat]]
[[Category:Comics featuring White Hat]]
[[Category:Artificial Intelligence]]

Talk:1872: Backup Batteries

2017-08-05T18:18:05Z

OriginalName:

2017-05-17T05:22:21Z

OriginalName: /* Transcript */

{{comic
| number = 1838
| date = May 17, 2017
| title = Machine Learning
| image = machine_learning.png
| titletext = The pile gets soaked with data and starts to get mushy over time, so it's technically recurrent.
}}

==Explanation==
{{This explanation may be incomplete or incorrect: Work in progress.}}

In this comic, Cueball explains to a Cueball-like guy his machine learning system. Cueball describes his machine learning system as a plumber would explain his plumbing. Data is input into a funnel, and goes through a mess of linear algebra, and comes out as answers. The main joke is that, despite this description being too vague and giving no intuition or details into the system, it is close to the level of understanding most machine learning experts have of the most popular class of techniques in machine learning, namely deep learning with neural networks (why reference to neural networks here? they are non-linear. better example is support vector machines.)

One of the most popular paradigms of machine learning is that of supervised learning, where a function mapping an input to an output is learned from several input,output pairs, e.g. a function mapping images of faces to people names, from a dataset of static labelled images. Classic machine learning techniques like regression, or logistic regression have understandable parameters, and provable algorithms, but require significant engineering in the pre-processing step and don't perform very well for data like images or natural text. Deep learning techniques on the other hand require very little pre-processing, but require the data to be run through several steps of linear algebra, where essentially in each step the output of the previous step is multiplied with a matrix and sent to the the next step. This multi-step process has proven to be very successful for image and text data, but the structure of the parameters, arranged as a matrix for each step allows for very little interpretation, and can only be described as "data going through a pile of linear algebra".

The method of training such deep neural networks is via gradient descent, which can be viewed as "stirring the pile of linear algebra until the answers start looking right".

The title text refers to recurrent neural networks, which are a useful class of deep neural networks for dealing with sequence data like speech or text.

This comic satirizes machine learning, more specifically neural networks. In it's most basic form, a neural network takes data and results and strengthens connections that give the right answer and weakens ones that don't, until the results "look right". Neural networks are extremely data dependent, and make remarkably few guarantees when compared to most other computing techniques, thus the joke.

Recently, other forms of neural networks, such as LSTMs, feed old sequence data back into the network with some delay, making it recurrent. The title text calls this the pile "getting mushy". The title text is also be a pun based on how Cueball is going through the data. Instead of using a shovel, he is using a canoe paddle. Canoes can be used on rivers, and rivers by definition have currents. Thus, a recurrent data could, in this situation, mean data treated as if it were part of a river.

==Transcript==
{{incomplete transcript|Do NOT delete this tag too soon.}}

[Cueball is standing on top of a pile of data with a funnel labeled "data" and box labeled "answers" in said pile]

Guy: This is your machine learning system?

Cueball: Yup! You pour the data into this big pile of linear algebra, then collect the answers on the other side.

Guy: What if the answers are wrong?

Cueball: Just stir the pile until they start looking right.

{{comic discussion}}

1838: Machine Learning

2017-05-17T05:19:29Z

OriginalName: /* Explanation */

{{comic
| number = 1838
| date = May 17, 2017
| title = Machine Learning
| image = machine_learning.png
| titletext = The pile gets soaked with data and starts to get mushy over time, so it's technically recurrent.
}}

==Explanation==
{{This explanation may be incomplete or incorrect: Work in progress.}}

In this comic, Cueball explains to a Cueball-like guy his machine learning system. Cueball describes his machine learning system as a plumber would explain his plumbing. Data is input into a funnel, and goes through a mess of linear algebra, and comes out as answers. The main joke is that, despite this description being too vague and giving no intuition or details into the system, it is close to the level of understanding most machine learning experts have of the most popular class of techniques in machine learning, namely deep learning with neural networks (why reference to neural networks here? they are non-linear. better example is support vector machines.)

One of the most popular paradigms of machine learning is that of supervised learning, where a function mapping an input to an output is learned from several input,output pairs, e.g. a function mapping images of faces to people names, from a dataset of static labelled images. Classic machine learning techniques like regression, or logistic regression have understandable parameters, and provable algorithms, but require significant engineering in the pre-processing step and don't perform very well for data like images or natural text. Deep learning techniques on the other hand require very little pre-processing, but require the data to be run through several steps of linear algebra, where essentially in each step the output of the previous step is multiplied with a matrix and sent to the the next step. This multi-step process has proven to be very successful for image and text data, but the structure of the parameters, arranged as a matrix for each step allows for very little interpretation, and can only be described as "data going through a pile of linear algebra".

The method of training such deep neural networks is via gradient descent, which can be viewed as "stirring the pile of linear algebra until the answers start looking right".

The title text refers to recurrent neural networks, which are a useful class of deep neural networks for dealing with sequence data like speech or text.

This comic satirizes machine learning, more specifically neural networks. In it's most basic form, a neural network takes data and results and strengthens connections that give the right answer and weakens ones that don't, until the results "look right". Neural networks are extremely data dependent, and make remarkably few guarantees when compared to most other computing techniques, thus the joke.

Recently, other forms of neural networks, such as LSTMs, feed old sequence data back into the network with some delay, making it recurrent. The title text calls this the pile "getting mushy". The title text is also be a pun based on how Cueball is going through the data. Instead of using a shovel, he is using a canoe paddle. Canoes can be used on rivers, and rivers by definition have currents. Thus, a recurrent data could, in this situation, mean data treated as if it were part of a river.

==Transcript==
{{incomplete transcript|Do NOT delete this tag too soon.}}

This is your machine learning system?

Yup! You pour the data into this big pile of linear algebra, then collect the answers on the other side.

What if the answers are wrong?

Just stir the pile until they start looking right.

{{comic discussion}}

1838: Machine Learning

2017-05-17T05:18:56Z

OriginalName: /* Explanation */

{{comic
| number = 1838
| date = May 17, 2017
| title = Machine Learning
| image = machine_learning.png
| titletext = The pile gets soaked with data and starts to get mushy over time, so it's technically recurrent.
}}

==Explanation==
{{This explanation may be incomplete or incorrect: Work in progress.}}

In this comic, Cueball explains to a Cueball-like guy his machine learning system. Cueball describes his machine learning system as a plumber would explain his plumbing. Data is input into a funnel, and goes through a mess of linear algebra, and comes out as answers. The main joke is that, despite this description being too vague and giving no intuition or details into the system, it is close to the level of understanding most machine learning experts have of the most popular class of techniques in machine learning, namely deep learning with neural networks (why reference to neural networks here? they are non-linear. better example is support vector machines.)

One of the most popular paradigms of machine learning is that of supervised learning, where a function mapping an input to an output is learned from several input,output pairs, e.g. a function mapping images of faces to people names, from a dataset of static labelled images. Classic machine learning techniques like regression, or logistic regression have understandable parameters, and provable algorithms, but require significant engineering in the pre-processing step and don't perform very well for data like images or natural text. Deep learning techniques on the other hand require very little pre-processing, but require the data to be run through several steps of linear algebra, where essentially in each step the output of the previous step is multiplied with a matrix and sent to the the next step. This multi-step process has proven to be very successful for image and text data, but the structure of the parameters, arranged as a matrix for each step allows for very little interpretation, and can only be described as "data going through a pile of linear algebra".

The method of training such deep neural networks is via gradient descent, which can be viewed as "stirring the pile of linear algebra until the answers start looking right".

The title text refers to recurrent neural networks, which are a useful class of deep neural networks for dealing with sequence data like speech or text.

This comic satirizes machine learning, more specifically neural networks. In it's most basic form, a neural network takes data and results and strengthens connections that give the right answer and weakens ones that don't, until the results "look right". Neural networks are extremely data dependent, and make remarkably few guarantees when compared to most other computing techniques, thus the joke.

Recently, other forms of neural networks, such as LSTMs, feed old sequence data back into the network with some delay, making it recurrent. The title text calls this the pile "getting mushy". The title text is also be a pun based on how Cueball is going through the data. Instead of using a shovel, he is using a canoe paddle. Canoes can be used on rivers{{citation needed}}, and rivers by definition have currents. Thus, a recurrent data could, in this situation, mean data treated as if it were part of a river.

==Transcript==
{{incomplete transcript|Do NOT delete this tag too soon.}}

This is your machine learning system?

Yup! You pour the data into this big pile of linear algebra, then collect the answers on the other side.

What if the answers are wrong?

Just stir the pile until they start looking right.

{{comic discussion}}