Data Science: Reality Doesn't Meet Expectations

danielfriedman | 462 points

> Moreover, you may quickly realize much of this work is repetitive and while time-consuming, is “easy”. In fact, most analyses involve a great deal of time to understand the data, clean it and organize it. You may spend a minimal amount of time doing the “fun” parts that data scientists think of: complex statistics, machine learning and experimentation with tangible results.

This. Universities and online challenges provide clean labeled data, and score on model performance. The real world will provide you... “real data” and score you (hopefully) by impact. Real data work requires much more than modeling. Understanding the data, the business and value you create are important.

As per #6, better data and model infrastructure is crucial in keeping the time spent on these activities manageable, but I do think they’re important parts of the job.

I’ve seen data science teams at other companies working for years on topics that never see production because they only saw modeling as their responsibility. Even the best data and infrastructure in the world won’t help if data scientists do not feel co-responsible for the realization of measurable value for their business.

Training integrative data professionals could be a great opportunity for bootcamps. Universities will (understandably) focus on the academically interesting topic of models, while companies will increasingly realize they need people with skills across the data value chain. I know I would be interested in such profiles. :)

hectormalot | 2 months ago

As a research-oriented data scientist at one of the larger tech companies, I can confirm that even here, a lot of people are unsure about what exactly data scientists are supposed to do. My most frequent request is "tell us why metric X dropped", to which the answer is often a subtle combination of many different factors (often random fluctuation) that doesn't lead to a pleasing actionable result in the sense of "here's why it dropped; go do this to fix it".

The really interesting research type work (Bayesian modeling, convolutional neural networks, etc.) takes a long time to implement and may produce no useful results, which is a really bad outcome at a company that measures performance in six month units of work and highly values scheduled deliverables and concrete impact. Many of the data scientists I work with tend to stick to methods that are actually quite simple (e.g., logistic regression, ARIMA) because these at least deliver something quickly, despite the fact that many of my coworkers come from research-heavy backgrounds.

In my org, there's nothing stopping anyone from pursuing advanced machine learning; for the most part we set our own agenda (in fact, determining priorities is part of the job role). And some people do in fact go after state-of-the-art ML, with some really cool results to show for it. But in terms of career progression and job safety, the risk is just way too high, at least for me personally. I save the highly mathematical stuff for a hobby.

Edit: while this may sound a bit negative, I will add that my description of data science isn't a complaint per se; I am mainly trying to inform those who are seeking a career in data science of what to expect compared to what is often promised. The work that is most valuable to a business is not exciting all of the time, but I don't think there is another job in the tech industry that I would find more enjoyable than my current one at the moment.

throwaway713 | 2 months ago

This article is pretty spot on. As someone who has worked in data science/analytics for over 6 years I have found that the field is filled with hype, managers who are not sure what data science actually is, and an absurdly wide amount of skills jobs expect you to be able to do well.

Apply for and interviewing for data science jobs is a total nightmare. You are competing against 100s or even 1000s of applicants for every job posting because someone said it was one of the sexiest careers of the 21st century. Further exacerbating this, Everyone believes that data is the new oil, and large profit multipliers are just waiting to be discovered in this virgin data that companies are sitting on. All that is missing is someone to run some neural network, or deep learning algo on it to discover the insights that nobody else can see.

The reality is that there is an army of people who know how to run these algos. MOOC's, blogs, youtube, etc have been teaching everyone how to use these python/R packages for years. The lucky few who get that coveted data science job can't wait to apply these libraries to the virgin data only to find that they have to do all kinda of data manipulating to make the algos even work, which takes days and weeks of mundane work. Finally they find out the data is so lacking that their deep learning model does very little in providing actual business value. It is overly complicated, computationally expensive, and in the back of your mind know you can get the same results using some simple logic.

Managers who don't understand data science fundamentals learn from the news and have their data scientist implement those buzz words so they can look good in front of their bosses.

I think there is a place for data scientists who understand the fundamentals of the models out there, and know when you should not use them. Data science is also increasingly a subset of software engineering and a good data science in a tech company should be able to code well. I also think that there is not some huge unmet demand for data scientists. Just a huge amount of hype and managers wanting to look good by saying they managed a data science team.

resolaibohp | 2 months ago

Data science is correctly valued when you realize how relatively unimportant it is. It is a small cog in a larger machinery (or at least it ought to be).

You see, decision-making involves (1) getting data, (2) summarizing and predicting, and (3) taking action. Continuous decision-making -- the kind that leads to impact -- involves doing this repeatedly in a principled fashion, which means creating a system around the decision process.

For systems thinkers, this is analogous to a feedback control loop which includes sensor measurements + filters, controllers and actuators.

(1) involves programmers/data engineers who have to create/manage/monitor data pipelines (that often break). This the sensor + filters part, which is ~40% of the system.

(2) involves data scientists creating a model that guides the decision-making process. This is the model of the controller (not even the controller itself!), which is ~20% of the system. Having the right model is great, but as most control engineers will tell you, even having the wrong model is not as terrible as most people think because the feedback loop is self-correcting. A good-enough model is all you need.

(3) involves business/front-line people who actually implement decisions in real-life. This is where impact is delivered. ~40% of the system. This is the controller + actuator part, which makes the decisions and carries them out.

Most data scientists think their value is in creating the most accurate model possible in Jupyter. This is nice, but in real-life not really that critical because the feedback-loop inherently moderates the error when deployed in a complex, stochastic environment. The right level of optimization would be to optimize the entire decision-making control feedback loop instead of just the small part that is "data science".

p.s. data scientists who have particularly low-impact are those who focus on producing once-off reports (like consultant reports). Reports are rarely read, and often forgotten. Real impact comes from continuous decision-making and implementing actions with feedback.

Source: practicing data scientist

wenc | 2 months ago

> I attended a 12-week data science bootcamp in mid-2016. ...

Yeah, well there's your problem, my dude. I've been doing what might be described as "data science" since I quit physics in 2004. Aka before the term existed. It's a great area to work in for intelligent people who want to use their brains to impact the real world; vastly better than what people get paid to do in physics. If customers don't know what the tools can do, it's because you as the data scientist have failed to explain it to the customer. If your work product isn't in front of the decision makers, you've also failed: they can tell the bottom line impact and will reward you accordingly. Sometimes there is no data in their data; they should know that up front.

As for whining about poor data quality: n00b. What do you think they're paying you for? Nobody gives a shit what people do in Kaggle competitions.

scottlocklin | 2 months ago

I stood up a data science operation at my company over the last few years, and have noticed a key difference in data-science projects that have been successful and those that have failed. It hits on a number of points brought up in the article, namely where does data science "fit" in an organization delivering software and how is the value realized by the business.

The worst cases I have seen is when executives take a problem and ask data scientists to "do some of that data science" on the problem, looking for trends, patterns, automating workflows, making recommendations, etc. This is high-level pie in the sky stuff that works well in pitch meetings and client meetings, but when it comes down to brass tacks this leaves very little vision of what is trying to be achieved and even less on a viable execution path.

More successful deployments have had a few items in common

1. A reasonably solid understanding of what the data could and couldn't do. What can we actually expect our data to achieve? What does it do well? What does it do poorly? Will we need to add other data sets? Propagate new data? How will we get or generate that data?

2. The business case or user problem was understood up front. In our most successful project, we saw users continuously miscategorized items on input and built a model to make recommendations. It greatly improved the efficacy of our ingested user data.

3. Break it into small chunks and wins. Promising a mega-model that will do all the things is never a good way to deliver aspirational data goals. Little model wins were celebrated regularly and we found homes and utility for those wins in our codebase along the way.

4. Make is accessible to other members of the company. We always ensure our models have an API that can be accessed by any other services in our ecosystem, so other feature teams can tap into data science work. There's a big difference between "I can run this model on my computer, let me output the results" and "this model can be called anywhere at any time."

While not exhaustive, a few solid fundamentals like the above I think align data science capabilities to business objectives and let the organization get "smarter" as time goes on as to what is possible and not possible.

danmostudco | 2 months ago

Teams being small, data being crummy, infra being hard, and yet expectations being high aren't so much complaints as the they are the job description.

The point of data scientists and the related roles listed in the article are not to just churn out the fun stuff, but to wade through the institutional and technical muck and mire it takes to bring the fun stuff to bear on a relevant business problem and to communicate the results in a way that people of all walks can understand.

kristjansson | 2 months ago

I'm generally confused by the hype around ML and 'data science'. it seems like CS has somehow regressed to the behavourism era of psychology or economics before the Lucas critique.

The problem with all this data talk isn't just about implementation or bad structure, the limitations of putting all your bets on inductive reasoning are systemic.

The insights that economists had in the 70s and 80s was that reasoning from aggregated quantities is extremely limited. Without understanding at a structural level the generators of your data, trying to create policy based on outputs is like trying to reason about inhabitants of a city by looking at light pollution from the sky.

My guess why data science so rarely delivers what it promises is because you can't get any value from historical data if your circumstances change to the point where past data is irrelevant. Which in the world of business happens pretty quickly. To have a competitive advantage, one needs to figure out what has not been seen yet.

And trying to exploit signals suffers from the issue laid out above. There was a funny case of an AI hiring startup trying to predict good applicants, and the result was people putting "Oxford" in their application in a font matching the background color

Barrin92 | 2 months ago

As a scientist, I've worked with data for decades. There's always been a prevailing belief that scientists and engineers with specialized domain knowledge are mostly fumbling in the dark and can be replaced with a general purpose technique.

This was certainly the vibe that I got from "design of experiments" when it was the statistical method du jour. Then from "Bayesian everything" and now "data science." I remember "design of experiments" studies being conducted with great fanfare and success theater, while producing zero results.

The long term theme is that science is hard for reasons that managers don't understand, can't manage, and are reluctant to reward.

analog31 | 2 months ago

I've seen a few similar articles now. Does this represent the general view of folks working in data science? "Data Science" is such as meaningless catch all term. The reality is in many organizations it's simply advanced business intelligence or advanced business analytics. There are some industries that lend themselves well to this whole practice, and they tend to be industries that have been borne out of the internet age (e.g. social media, internet advertising, etc.)

Some other industries have been doing "data science" for ages. Credit Risk Modelling, insurance and so on.

Every time I read one of these articles, I feel it's just an individual who entered a kind of crummy situation and they're learning what it means to work in a corporate environment. Some are better than others. Some are more motivated than others. Some have better cultures than others. Some are more willing to make technology a key part of their business strategy. Some are more data driven than others.

My recommendation is to always ask the fundamental question before joining: what are you trying to achieve with data science, and is it actually achievable?

rafiki6 | 2 months ago

As a data dude in public/nonprofit healthcare-landia I agree with all this, plus:

- It's essential to have/develop domain expertise in your industry.

- Beware plausible, but incorrect (or poorly interpreted) data that supports yours (or others') assumptions/biases.

- Add on to #4 - at least as bad as this is having well-intentioned people on your team who "know enough (a bit of SQL or low/no-code data tool") to be dangerous. Um, why are you joining unnecessary tables, or using a different alias for the same columns/tables in different queries, with no comments or standard formatting?

- Hold your nose, but anything you do in SQL/R/Python/even fancier programming tool/language is going to pass through MS Excel at least once sooner or later which can irreversibly bastardize CSVs (even just opening without saving!), truncate precision to 15 digits, change data types, etc.

- So glad for the callout in #7 - there are clearly devs/data folks out there who are happy to take on an "interesting programming project at a great paying job" - that isn't serving the best interests of humanity.

Optimal_Persona | 2 months ago

This rings very true to me. I'm working on moving over to an SWE role in the next few years for many of these reasons.

I'll just add one: the business absolutely doesn't care how you get your answer, only if they're reliable enough (hand grenade close is better than most companies have today).

While this seems obvious enough to anyone with a few years under their belt, to the new DS grad who has their time series analysis canned in favor of slapping a simple moving average in place and shipping it can be rather disillusioning.

Icathian | 2 months ago

I’ve been doing an MS in Data Science very slowly due to work and 2 new kids. Finishing the degree this year in year 4. I was very excited about the prospect of doing something different. A few things have changed for me.

1). I am hearing about Data Science Teams being furloughed during these times. That isn’t happening in my function (Corporate Finance). I am glad to be secure even though I enjoy much of the data sci work.

2) I’m able to apply Data Science concepts in my current role, and it’s adding a lot of job security and providing me with exposure. I am much less interested now in moving to straight Data Science and instead am applying my learnings in my current role as a sort of in-house Data Science guy. But I have a lot to learn to be honest.

3). There seem to be a lot of “thought leaders” acting like they are big experts in the area and really don’t know anything many of us amateur scientists don’t know. They pull perfect clean datasets and show these magic transformations they just copy from others to get YouTube hits or Twitter followers. That just never happens in real life, and many leaders are seeing this and losing interest in this function in the returns they are getting from sole data science folks.

Vaslo | 2 months ago

I work as a data scientist. Some of the author's points are workplace-specific: lack of leadership, being the only data person, ethical concerns. The others are just aspects of the job - communicating about your job and impact, dealing with vague specs or managing low-quality datasets.

Neither of those quite match the articles title, perhaps it just refers to the author's personal expectations. Neither of them seem that specific to data science, or without parallels in other software jobs. And neither of the points read like a slight towards data science to me, like some of the other commenters here suggest.

s1t5 | 2 months ago

One issue might be that organizations subconsciously resist the data scientist, or more generally, the nerd in his/her attempt to take over decisions. If these decisions are invariably tied to the goals and careers of managers, how can the data scientist have a "seat at the table" in all but the most enlightened and technical companies? The disorganized state of data and infrastructure suits the ambitious manager well, who can just put in enough effort to find data to have their project greenlightened or to answer one specific question.

Progress may only come slowly, ideally through products bought from 3rd parties whose results are understood and controlled by management.

UweSchmidt | 2 months ago

I did "data science" for about a decade, consulting with plaintiffs firms and state AGs on antitrust and fraud cases. For each case, the work flow was roughly this:

-- write discovery requests

-- review production, and check out data and documentation

-- write supplementary discovery requests

-- review production, and check out data and documentation

[repeat as needed]

-- analyze data, and write deposition questions

-- help attorneys wring answers from deponents

[repeat as needed]

-- analyze data, and produce required output

-- write parts of briefs and expert reports

I generally did that in consultation with testimonial experts and their data analysts. Sometimes that didn't happen until we'd documented the case enough to know that it was worth it. And occasionally small cases settled with just me as the "expert".

It's a small industry, and not easy to get into, unless you know key players at key firms. But the money's pretty good, and the work can be exciting. I loved being that guy in depositions whispering questions to the attorneys :)

This all involved pretty simple calculation of damages, through comparing what actually happened vs what would have happened but for the illegal behavior. But-for models were typically based on benchmarks.

After data cleanup in UltraEdit, I did most of the analysis in SQL Server. I used Excel for charting and final calculations.

mirimir | 2 months ago

This reads like Indiana Jones teaching Archeology. Yes, as a data-scientist you actually have to work, most of the work is digging in dirt, and mostly you won't find anything of interest.

avip | 2 months ago

It works well when subject matter experts exist in the org and collaborate/supervise/drive data folk, to solve some issue the sme's have spent enough of their own time thinking about.

If its just data folk by themselves getting dumped with org data and told to find pirate gold...then its a crap shoot.

op03 | 2 months ago

The real issue with data science, from the perspective of ML pipelines/using ml in products, is most people are straight up not smart enough for it. The second the problem falls outside the bounds of a commonly used model, 90% of data scientists are ill equipped to come up with a profitable solution. So they stumble around in the dark, producing nothing of real value. People underestimate the degree to which extreme mathematical maturity and skill can bend the results of commonly used ml models.

codingslave | 2 months ago

This reads as a series of bad job experiences and I think is explained by a wide variety of job functions that all can have "Data Scientist" as a title. Someone else's experience could be totally different. You have to know what to look for and what to avoid. If you're trying to find a DS job, one of your top priorities is finding out what the actual job consists of. For instance, a Data Scientist at Facebook might be called a Data Analyst at many other places--no modeling required.

I know this because I've been on that journey. But there's no reason to expect some department head that's never been exposed to DS to know this. They just copy/paste some other company's job req. If you're more junior, here are my tips:

- If it's a "new DS team" that supports a variety of teams: beware. Bolt-on DS doesn't work well, as it's really hard to build a meaningful solution that's not deeply integrated.

- If it's an old company or in a conservative industry: beware. There are likely to be data silos and difficult ownership models that make it nearly impossible to get and join the data you need.

- If it's a small company: beware. You're likely going to need a broad set of knowledge that's won with several years of experience to be able to build end-to-end solutions that are integrated into the rest of the tech stack.

- If it's not an engineering-driven culture: beware. DS will often be used to provide evidence to someone else whose already made up their mind and pretend they're being data-driven, and you'll be the disrespected nerd that's expected to do what it takes to deliver the answer they want. Most companies claim to be "data-driven", few are, and even fewer understand data-driven isn't always possible or desirable.

Industry is still trying to figure out how to use ML and are still learning that it's not as easy as hiring someone that knows about all the algorithms, but rather it takes deep technological changes to data infrastructure to enable the datasets that can then be used by the ML experts. But you don't have to be the person that helps them figure this out the hard way (i.e. by being paid to not accomplish much due to problems outside of your control). Better to find a place with a healthy data science team that can help you learn and contribute. They exist.

agentofoblivion | 2 months ago

Great read. A lot of those problems are real, and some of those I’ve experienced myself. But I think at least some of them are related to the immaturity of the field. We’re only at the beginning of creating the tools and platforms to facilitate DS, making it more reproducible and easier to measure.

For example, I’m working on the tool to make data management easier and convert datasets into a structured representation. If you have experienced that you spend a lot of time on preparing and analyzing data, and it is tedious, please reach out to me michael at, would love to get your feedback on the product we have built so far.

deppp | 2 months ago

A really easy way that I try to explain things to people is like this:

You can't compress information until you have it in a format that is appropriate for compression.

That is:

You can't compress (apply/create algorithms) information (data) until you have it (instrumented data collection) in a format (schema) that is appropriate for efficient compression (structured logging/cleaning).

99% of that is Data Engineering and building good engineering practices which have good data practices as a priority.

For any organization that has more than a handful of employees and more than one product, that is a non trivial task and gets more difficult the larger the organization gets.

AndrewKemendo | 2 months ago

I wrote a blog post along similar lines in 2018 ( ); unfortunately, the industry hasn't changed much since then.

As noted in the submission, there's a lot of flexibility in what a "data scientist" is. Normally that's good and healthy for the industry. However, it contradicts a lot of optimistic bootcamps/Medium/YouTube videos, and many won't be prepared for the difference.

minimaxir | 2 months ago

I've been a data person for the past year and a half and I'm very disappointed with the bewildering array of titles out there and the rather vague meanings behind them (Data Analyst, Data Scientist, Data Engineer, ML Engineer).

It's overall hurting my ability to build my personal brand and seek roles that are a fit for my existing skillset and aspirations.

What exactly does 'ML Engineer' communicate to employers in terms of baseline skills? Is the role closer to that of a data engineer or an analyst?

stuxnet79 | 2 months ago

From my perspective as a data person, everything on this list is true. I would add on to #4 to say "You're likely the only data person" to say "You're likely the only data person and expected to do everything you need to do your job yourself" (from sourcing the data to deploying your model).

High Data Scientist salaries and expectations combined with a shortage of qualified people often mean you're expected to be a one-person band, which I find to be miserable.

arafa | 2 months ago

Point 5, “ Your impact is tough to measure” is also shared by Quality Engineering and SRE, and not unique to Data Science. The point about being a support role holds true for them and it is thoroughly frustrating when a front-end dev makes a small change to a visual element is praised to the roof while complex automation projects by the quality team, ingenious recovery and reliability projects by SRE, and massive and fascinating inferences by data science are undervalued by leadership. The truth is most leaders just can’t connect the dots. I’ve worked as a full stack engineer btw do not taking a dig at front-end work, but it’s clearly easier to measure impact. I’ve worked in quality too and when you’re only called in to ask why one bug got out and never asked about the thousands you’ve stopped it’s demoralizing. It’s part of the reason I started Tesults (, if you’re in one of these support roles, measure, measure and measure and throw those reports into the faces of leadership. It shouldn’t have to be done but without it, the point the author is making here will take place.

ajeet_dhaliwal | 2 months ago

Coase: "If you torture the data long enough, it will confess to anything."


smitty1e | 2 months ago

I guess I'm in the minority in these threads..? I've been doing machine learning / model-building / pushing models to prod and maintaining for about 6 years now. It's still 50/50 understanding the data and building/tweaking/training/testing models. But it sounds like most people with this title are analysts? At least that's what posts and threads lead me to believe. I've also met a lot of people with titles like "ML Engineer" or "Data Scientist" who don't do machine learning. They are analysts, engineers, or maintaining data pipelines.

eanzenberg | 2 months ago

> You’re likely the only “data person... Because people don’t know what data science does, you may have to support yourself with work in devops, software engineering, data engineering, etc.

Nothing has summed up my entire working experience more than this, it’s almost painfully accurate.

On one hand it’s an exciting challenge, you learn a lot and you get good at adapting to these situations.

On the downside I have practically no senior data science people to turn to for help when I do need it, which is frustrating.

FridgeSeal | 2 months ago

I am sorry to sound like I am being obstinate, but my opinion about this is that as a society, since the early 90s, we have put way to much focus on "tech" than we have put on plain old mathematics or foundational science.

I don't mean manufacturing (which is doing really well), but companies like Microsoft, Google, Facebook (and even Apple) and others do encourage you to try to compete against their founders (or maybe society does that) rather than focusing on being solid mathematically. Yes, Google pays people well with those skills, but movies portray mostly their founders, emphasising how rich they are, while mathematicians are generally portrayed as weird. Society as a whole puts more emphasis on Bill Gates than on fundamental researchers.

In fact, if you really want to have a rich representative, you can pick the Simons guy. (See, I don't even know his name.) His Medallion hedge fund was built on mathematics. Ironically, Bill Gates is these days one of the biggest financial supporters of people with science skills that he doesn't have.

It is a fad to be a techie. Mathematics is not a fad, although it does have internal fads.

mikorym | 2 months ago

I do not understand. Have never understood. "Data Science" is, surly, newspeak. The appropriate term, surly, is "statistics".

worik | 2 months ago

I worked as a data scientist for 4 months at a VC firm. I have a PhD and thought the work might be legit when I was hired. After the 4 months I quit when it became apparent that my credentials were being used for managerial intrigue and the work was essentially a joke, with no rigor at all. This article hits the nail on the head, unfortunately, these positions are not often real jobs.

erdos4d | 2 months ago

There is a philosophical principle which says that any model superimposed on reality could be seen as reality itself, while it is merely a superimposed interpretation, in principle.

Korzybski formulated these principles, among other things.

Most of data science models are as wrong as astrology and numerology. They have no connection to reality, or rather inadequate.

This principle explains abysmal failures of all Model-based "sciences", stating from financial markets and up to virus spreading models.

Simulations of non-discrete, non-fully-observable (AI terminology) system has exactly the same relationships with underlying reality as a Disney cartoon to a real world.

This is why expectations will never be meet, except for natural (non-inaginary) pattern recognition.

A drop of proper philosophy worth years of virtue signalling.

johndoe42377 | 2 months ago

> internally, you can make inroads supporting stakeholders with evidence for their decisions!

The problem is that this can all too easily become motivated reasoning: one provides a stakeholder with support for the decision he already made. From his point of view, this is a valuable service, but it does the organisation a disservice: decisions should be made after considering the data, rather than consider only those data which support a decision.

Also, while ethical issues certainly arise, I think that Greyball is not a good example. Uber evading police enforcing the taxi monopoly is no more unethical than the Underground Railroad evading fugitive-slave agents. The taxi monopoly is itself unethical, and evading it increases the common good.

zeveb | 2 months ago

Disclaimer: I use the term Data Scientist throughout this post; however, popular titles such as Data Analyst, Data Engineers and BI analyst are randomly applied by people who know nothing, and these people share none of the responsibilities of a Data Scientist.

I have never had hopes about the potential impact of being a Data Scientist. I felt every company should be a “data company”, but everything I knew told me that companies are political institutions bounded by the pressures of late stage capitalism. Anyone who things different is dim, anyone who blogs about it is a moron.

My expectations did meet reality.

Where did my expectations come from?

I attended a four year Computer Science degree, followed by four and a half years of earning a Ph.D. I then spent 20 years in industry. 19 of the 20 weeks’ focus were not on machine learning (ML) and artificial intelligence (AI).

I figured I’d spend most of my time buried in code and data, I was right, I had to find shit buried in it, and dig it out with my teeth. Executives hated me because I was a threat, but they needed me so I continued to get paid. I continue to be able to create insight and predictions that almost no one else can, and until this stops I will get a 200k a year salary, benefits and a Tesla.

All of this happened, I can't be bothered to waste my time commenting on this moronic blog post.

sgt101 | 2 months ago

Work reality rarely meets expectations. I am sure a lot of “UX” people also got a little disappointed that reality is much more mundane than the fancy title suggests. Or the people whose last algorithm work was the interview.

Ididntdothis | 2 months ago

It really depends on the niche or the industry considered, though: I can happily say that I can do materials informatics from my basement at home now and much faster and better than as a cog in any lab anywhere in the world. Same for a great number of STEM applications, if you ask or follow high-level practitioners through conferences, journals and social media. The elephant in the room is the Intellectual Property generated through STEM applied data science, which is hot and even dangerous as you can see from superstars like OpenAI, DeepMind or politically-motivated aggregations.

DrNuke | 2 months ago

The most common complaint I've heard from the data science team is that there isn't enough data to work with.

I'm not fully convinced that data science with ML and more modern techniques are applicable across domains out of the box. I think there is value to be added if data scientists can specialise in domains.

If we take humans as an analogy, even with the kind of general intelligence we have, we need domain expertise to be able to have advanced intuitions and make predictions about the future. I believe this is true for data science as well.

kovac | 2 months ago

I was a data scientist for one year. Experienced many of the adverse situations explained in the article, plus I thought it isn't for me. I joined my next job as a software engineer (after an extensive interview prep). Couldn't be happier. Still doing plenty of data science. But my product is actually a product, not the analysis (as is often the case with DS). I feel "more central" to the project, to the company. I'm still building ML models, features etc for a living.

starchild_3001 | 2 months ago

As the lead data scientist at a small-ish fintech, I can confirm many of the frustrations and disappointments in the OP. But my trajectory was slightly different - from being the only "data science guy" in 2016, to now leading an autonomous team of four, with quarterly meetings with the CEO, and monthly meetings with our tech leadership. I decide tech stack, workflow, and hiring. Execs decide priorities. Sure, some of it was dumb luck, some of it was actually having a CEO that cares about data strategy, but I like to think at least some of it was me.

So here's what I think I did right:

1. Provide indisputable, obvious business value every month. You should consider yourself an in-house consultant to whichever cost center your salary is drawn from. If you're product development, prove value to them. If you're operations, or sales, or marketing, prove value to them. After about two months, you should be able to justify your existence in two sentences. Just remember, most of your company probably thinks of you as a optional add-on.

Your first few projects should attack high-impact pain points with the simplest solutions possible. My first projects were basically ETL into some basic regression into a dashboard. No machine learning required. But it was better then what they had (which was often nothing), and it was STABLE and RELIABLE. And that leads to the next point...

2. Build trust. With my dead-simple models, nothing ever blew up, there were no nonsensical answers, and there wasn't much brittleness when new categorical features or more cardinality was added. It mostly just worked. And that built my reputation for me. They didn't have to understand what was going on in the model, but they knew, from experience, that they could trust the result. Once I had the credibility, I could start building more complex, more elaborate models, and asked them to trust those as well. If they don't trust your models, then no business value has been created, and your job is worthless.

3. Recognize that data science is being done everywhere in the organization, and respect it. Every department has someone who has built a monster spreadsheet that contains more embedded domain knowledge then you could hope to learn in a month. As data scientists, we like to think that we're helping the organization by building critical metrics to improve performance. But here's the catch. If the metric was truly critical, someone has built it already. It might be ad-hoc, use poor-methodology, and be somewhat wrong, but it works and is good enough. You have to find that person, learn from them, and improve on it.

4. Be as self-contained as possible. Ideally, your critical path should not depend on other teams doing things for you (except for IT setting up data access). You should be able to do it all. From front-end dashboards, to ETL, to DevOps. Remember, you're an in-house consultancy. You should be able to take problems and just handle them, rather then be a perpetual bother and distraction to other teams.

There's more, but if you do these four things, I think you can build the reputation in your company for creating useful, accurate data tools that help other people do their jobs better. After that's achieved, people will breaking down your door to get your help. That's where my team is now - we've got a backlog for at least 18 months, with our work priorities often being set directly by the CEO.

StonyRhetoric | 2 months ago

My feeling is that a lot of companies think: "We need a data scientist because all the big players also have one!"

In fact, they actually don't need a data scientist. At best they need someone who cleans data, creates pie charts or even worse, they relabel the database admin job as "Data scientist".

simonkafan | 2 months ago

Can definitely relate to this. Work for big consulting firm (F500) as a data scientist, end up in this weird software engineer/ml engineer hybrid role.

I personally love it but am doing more pure software engineering now as the infrastructure is not there and I need to build it myself.

Rainymood | 2 months ago

To point #5 in the article, in my experience, ascending order of potential to generate value for business:

An astonishingly large fraction of Data Science output goes to die in pretty presentations.

From what's left, a large fraction ends up in Spreadsheets.

A disappointingly small fraction ends up in live services.

alixedi | 2 months ago

Completely agree. I made a post couple of weeks ago trying to find some solutions for this:

new_learner | 2 months ago

it's a problem with tech in general. some things come over-hyped. and in the process people forget what's the actual problem to be solved because they fell in love with tools | tech. maybe the solution could easily be done in excel but then that's not sexy. I personally prefer to handle most parts in Python because of automation. writing functions in python is easier than writing functions in SQL or Excel(macros)

dzonga | 2 months ago

Maybe I have a narrow set of experience, but in my mind a “data engineer” is not a substitute for a “data scientist”.

mjparrott | 2 months ago

The situation sounds similar to ones years ago for statistics, operations research, optimization, and management science.

I view all of such work as applied math.

My experience is that applied math, from the fields I mentioned and some more recent ones, and more, with emphasis on the more, can be valuable and result in attention, usage, and maybe money.

I've had such good results and have seen more by others.

Some examples:

(1) Airline fleet scheduling and crew scheduling long were important, taken seriously, pursued heavily, with results visible and wanted all the way up to the C-suite.

(2) Similarly for optimization for operating oil refineries: So, here is the inventory of the crude oil inputs and the prices of the possible outputs. Now what outputs to make? The first cut, decades ago, was linear programming, and IBM sold some big blue boxes for that. More recently the work has been nonlinear programming.

(3) The rumors are, and I believe some of them, that linear programming is just accepted, used everyday, in mixing animal feed.

No surprise and common enough, IMHO what really talks is money. If can save significant bucks and clearly demonstrate that, then can be taken seriously.

But from 50,000 feet up, tough to get rich saving money for others. If they have a $100 million project and you save them $10 million, then maybe you will get a raise.

What's better, quite generally in US careers, is to start, own, and run a successful business. If that business is to supply the results of some applied math, and the results pass the KFC test, "finger lick'n good", then charge what the work is worth.

Maybe now Internet ad targeting is an example.

I'm doing a startup, a Web site. The crucial enabling core of what I'm doing has some advanced pure math and some applied math I derived. Users won't be aware of anything mathematical. But if users really like the site, then it will be mostly because of the math. So, it's some math -- not really statistics, operations research, optimization, machine learning, artificial intelligence, or management science -- it's just some math. The research libraries have rows and rows of racks of math; I'm using some of it and have derived some more.

Generally I found that the best customer for math is US national security, especially near DC. E.g., now some people are building models to predict the growth of COVID-19. Likely the core of that work is continuous time, discrete state space Markov processes, maybe subordinated to Poisson processes. Okay: One of the military projects I did was to evaluate the survivability of the US SSBN (ballistic missile firing submarines) under a special scenario of global nuclear war limited to sea -- a continuous time, discrete state space Markov process subordinated to a Poisson process. Another project was to measure the power spectra of ocean waves and, then, generate sample paths with that power spectrum -- for some submarines. There was some more applied math in nonlinear game theory of nuclear war.

Here's some applied math, curiously also related to the COVID-19 pandemic: Predict revenue for FedEx. So, for time t, let y(t) be the revenue per day at time t. Let b be the total market. Assume growth via virality, i.e., word of mouth advertising from current customers communicating with remaining target customers. So, ..., get the simple first order differential equation, for some k,

y'(t) = k y(t) (b - y(t))

where the solution is the logistic curve which can also be applied to make predictions for epidemics. This little puppy pleased the FedEx BoD and saved the company. Now, what was that, data science, AI, ML, OR, MS, optimization? Nope -- just some applied math.

I have high hopes for the importance, relevance, power, fortunes from applied math, but can't pick good applications like apples from a three.

graycat | 2 months ago

tldr OP has been raised by television to believe that one day he'd be millionaires and movie gods and rock stars and data science gurus, but he won't and he’s slowly learning that fact. and he’s very very pissed off.

you mean you didn’t realize “data scientist” is just a sexed up “statistician”?

sorry you got hyped into shelling out for some bootcamp...

tibyat | 2 months ago