Why hasn't LLT hired a Data Scientist yet?

[Haris Javed]

Weird but important question. Why hasn't LTT hired a data scientist yet? a data scientist can play a huge role in many aspects of the LTT channel. E.g subscriber prediction, viewer engagement with LTT content, network traffic, anomaly detection. user engagement with the LTT website - just to name a few 

 

Thank you

 

[Origami Cactus]

My theory would be that they are still so small that they don't need one, as youtube analytics and site analytics exist(for free), and those give enough info for a channel their size.

[Stormseeker9]

6 minutes ago, Haris Javed said:

Weird but important question. Why hasn't LTT hired a data scientist yet? a data scientist can play a huge role in many aspects of the LTT channel. E.g subscriber prediction, viewer engagement with LTT content, network traffic, anomaly detection. user engagement with the LTT website - just to name a few 

 

Thank you

 

Why important? Important to you?

 

I’m sure the Linus media group are capable of running a business like they are now..

[Guest]

Colton probably does that with Sales, Marketing and being fantastic. 

[dizmo]

I highly doubt they'd be big enough to warrant such a position.

[191x7]

Imagine them having a "GDPR Officer" although they aren't from the EU? loooooooooool

[Haris Javed]

11 minutes ago, Stormseeker9 said:

Why important? Important to you?

 

I’m sure the Linus media group are capable of running a business like they are now..

hmm, I guess you can say its important to me, but it should be important to any business. Would you not be interested to find out how users engage with your website so you can make your site easier to navigate? Would you not be interested to find out that you lost 35% of your subscribers in Canada, and gained 50% new subscribers in Australia? How would you adjust your subscription models now that your subscriber base has shifted? and also predict what would be the best time to release the video to get the most user engagement with it? What if you could predict network failures before the failure takes place and brings down the network?  I do feel that LTT is large enough to benefit from a data scientist. 

 

Thank you

[Origami Cactus]

10 minutes ago, Haris Javed said:

Would you not be interested to find out how users engage with your website so you can make your site easier to navigate?

 

Would you not be interested to find out that you lost 35% of your subscribers in Canada, and gained 50% new subscribers in Australia?

 

How would you adjust your subscription models now that your subscriber base has shifted?

 

and also predict what would be the best time to release the video to get the most user engagement with it?

All of those are easily doable with Youtube/Google analytics, so Colton already does all that.

 

[Haris Javed]

Just now, Origami Cactus said:

All of those are easily doable with youtube analytics, so Colton already does all that.

 

 

ok

Hi, my question was out of curiosity but I will share some more info

 

I use google analytics. Its great for data visualization, but has 0 predictive power. 

google analytics gives you an aggregate traffic view, so it will tell you that you have 10K visits to a page / site, but has limited view into unique user information. (This can be fixed if a tool like aws kinesis is used) 

It cannot predict user behavior e.g why user left a page? were they frustrated? did they fail to find what they were looking for? 

aggregation hides the lack of accuracy - this might not be a big deal because we are not dealing with Medical / Patient data 

 

just my thoughts

 

Thank you

 

 

[NineEyeRon]

23 minutes ago, 191x7 said:

Imagine them having a "GDPR Officer" although they aren't from the EU? loooooooooool

Data Protection Officer and they probably don't need one.

 

Even with financial data its not mandatory, unless your handling lots and lots and doing weird things with it.

[Guest]

42 minutes ago, Haris Javed said:

hmm, I guess you can say its important to me, but it should be important to any business. Would you not be interested to find out how users engage with your website so you can make your site easier to navigate?

They click on a topic to read it. It's not exactly hard to navigate. LTT's audience are tech savvy, that's why we watch it. 

42 minutes ago, Haris Javed said:

Would you not be interested to find out that you lost 35% of your subscribers in Canada, and gained 50% new subscribers in Australia?

Those kinds of fluctuations don't happen with a large channel. 

42 minutes ago, Haris Javed said:

How would you adjust your subscription models now that your subscriber base has shifted?

I wouldn't. It's still an increase. 

42 minutes ago, Haris Javed said:

and also predict what would be the best time to release the video to get the most user engagement with it?

Daily seems to work well. User engagement is for the author, not the audience. 

42 minutes ago, Haris Javed said:

What if you could predict network failures before the failure takes place and brings down the network?

What if you had a redundancies in place, like LTT does. 

42 minutes ago, Haris Javed said:

 I do feel that LTT is large enough to benefit from a data scientist. 

Other employees are fully qualified to take out that role, and probably do. Most of it is just common sense. Sorry if I'm a bit frank but I fully believe consultants are just a pyramid scheme 100%. 

 

[Haris Javed]

Hey no worries, I am not a consultant, I am not looking for a side gig either, but I asked the question out of curiosity. 

 

They click on a topic to read it. It's not exactly hard to navigate. LTT's audience are tech savvy, that's why we watch it. 

- What can I say, People like to think they know everything. Yes navigation is not hard, but my question was more focused towards the insights

 

Those kinds of fluctuations don't happen with a large channel. 

- That was hypothetical scenario. You say those large changes dont happen, however on July/27/2016 LTT had anomalous  loss of -16.5K subscribers, and gained 13.9k the next day, similar anomaly around Sept /20/2018, and December/13 /2018

 

I wouldn't. It's still an increase. 

- fair point, but you can still see  increasing traffic with a decreasing rate, think calculus

 

Daily seems to work well. User engagement is for the author, not the audience. 

- fair point

 

What if you had a redundancies in place, like LTT does. 

- businesses much larger than LTT run into this issue. Companies like Boeing, 3M, Cargil, Target (corporate giants, with insane redundancy) can / do go down. This is why they use Machine Learning for load prediction. In this case you can say that LTT is small enough to not be impacted by downtime and because real time data is not important

 

Other employees are fully qualified to take out that role, and probably do. Most of it is just common sense. Sorry if I'm a bit frank but I fully believe consultants are just a pyramid scheme 100%

- Nothing against LTT employees. They are great people , and produce some of the most fun content to watch. However if things were as easy as using your common sense, you wouldn't need specialized doctors (nurses can tell if a patient is having a heart attack or if a patient is going in a Diabetic Ketoacidosis, but they don't perform angioplasty or start a Ketoacidosis treatment), most people realize that littering is bad for the environment but a scientist can explain to you what economic / environmental impacts it can cause. My point is that experts in a domain exist for a reason. Many data scientist are not consultants, they provide answers to real world problems, and most of them do it because they love what they do. Their goal is not to flaunt around their PhD's but to use their their knowledge to better humanity. If you ever get a chance go to Kaggle.com, check out how many real world problems have been solved by data scientists. You like Alexa? Thank a data scientist for that, You like Image recognition on your phone? Thank a data scientist for that, You like Binge watching netflix? thank a DS for that. You like early Cancer detection? Thank a DS for that, You want to know if your heart is healthy just by listening to its rhythm? Thank a DS for that - all this work was done free.  

Edited March 26, 2019 by Haris Javed
A grammar error was triggering forum users, which is more important than the content being discussed

[LogicalDrm]

I see problem, and probably reason why they haven't hired anyone special. Also why they did start FP. Its called Youtube. When you are on platform which doesn't publish info about how their algorithms work, and which they tweak and change all the time, its vastly different thing than with predicting disease spreads or stock markets. Both of those have a lot of public data and end result is made out of many variables. Not with Youtube. Or rather you have no idea what variables would be and equation is changed without giving you any info why or when. Hiring someone for that kind of work is kinda wasted money when you could hire someone actually capable of making content better. Like producer or another marketing expert, maybe someone who has experience from TV or printed media.

[Haris Javed]

6 minutes ago, LoGiCalDrm said:

I see problem, and probably reason why they haven't hired anyone special. Also why they did start FP. Its called Youtube. When you are on platform which doesn't publish info about how their algorithms work, and which they tweak and change all the time, its vastly different thing than with predicting disease spreads or stock markets. Both of those have a lot of public data and end result is made out of many variables. Not with Youtube. Or rather you have no idea what variables would be and equation is changed without giving you any info why or when. Hiring someone for that kind of work is kinda wasted money when you could hire someone actually capable of making content better. Like producer or another marketing expert, maybe someone who has experience from TV or printed media.

Thank you, this is the first response that makes sense to me, since I am not familiar with how internals of YouTube work.  The issue is that we have no way to  access user data to test and train a model . 

[Spotty]

4 hours ago, Haris Javed said:

If you ever get a chance go to Kaggle.com, check out how many real world problems have been solved by data scientists without any monetary compensation. You like Alexa? Thank a data scientist for that, You like Image recognition on your phone? Thank a data scientist for that, You like Binge watching netflix? thank a DS for that. You like early Cancer detection? Thank a DS for that, You want to know if your heart is healthy just by listening to its rhythm? Thank a DS for that - all this work was done free.  

For people who are meant to be smart, data scientists are really stupid.

[Haris Javed]

1 hour ago, Spotty said:

For people who are meant to be smart, data scientists are really stupid.

Sure, why not. After a long workday my brain tends to get tired, and inevitably incoherent nonsense comes out. My question has been answered by logicaldrm but your comment added nothing beneficial  - You went over my response and all you could do was show your ignorance while missing the point of the discussion. The point was that data scientists from different domains have provided free solutions to practical/real world problems. I am guessing you are tons of fun at office meetings. Keep it up

[kasugatei]

A bit off topic, but I am curious. Why majority of the people which are doing data science must behave like Jehovah's Witnesses and spread "the gospel of data science" to the "common people" and propose that data science is solution for everything?

 

Also on topic. Hiring data scientist would probably cost more money than the gain LTT would have for having one. They are a way too small company for needing one. Someone with a bit better knowledge of the statistics is more than enough for them and, as some people have already mentioned before, they already have people doing that. Youtube provides more than enough data and it is not massive amount of data that you can't process manually.

 

To conclude, data science and machine learning have their place in the world, but this is not one of them.

[Tech_Dreamer]

i bet they could hire another employee for the overall benefit of the channel  , but they're probably keeping costs low & going with the flow. More catchy thumbnails is probably their settled science at this point having a certain amount of views each time for each video.

[Haris Javed]

18 minutes ago, Niksa said:

A bit off topic, but I am curious. Why majority of the people which are doing data science must behave like Jehovah's Witnesses and spread "the gospel of data science" to the "common people" and propose that data science is solution for everything?

 

Also on topic. Hiring data scientist would probably cost more money than the gain LTT would have for having one. They are a way too small company for needing one. Someone with a bit better knowledge of the statistics is more than enough for them and, as some people have already mentioned before, they already have people doing that. Youtube provides more than enough data and it is not massive amount of data that you can't process manually.

 

To conclude, data science and machine learning have their place in the world, but this is not one of them.

To conclude, data science and machine learning have their place in the world, but this is not one of them.

- Yes i agree - this is not the place, I would consider my question answered / explored. 

 

A bit off topic, but I am curious. Why majority of the people which are doing data science must behave like Jehovah's Witnesses and spread "the gospel of data science" to the "common people" and propose that data science is solution for everything?

 

- I guess its curiosity, and how maximum amount of information can be extracted from something, but I will not post any DS related questions on LTT.  Instead of a constructive discussion, all I got was "shut up", "You are stupid"