I'm not one of those people hung up on traditional college education, I got my degree mostly studying by myself and always found most of what my college offered useless, so I'm not coming from the angle of "you need an x year college degree for this...".
I realize this is an harsh opinion but it's not cool to get north of $10k from people that need a job and tell them "we'll make you a designer in 3 months", that just doesn't happen.
I’ve personally interviewed around 25 GA grads, and the result was pretty much the same with all of them - “great, you now know what you don’t know, and have about a year of self-directed learning followed by a year of on the job experience before being employable in a role above an internship.
It’s actually pretty heartbreaking when you talk them and get a sense for how much they were oversold.
My heart goes out to those people, and I'm guessing Web Development Immersive students?
I always described GA's programs as a multiplier in that your outcome is going to be weighted by your experience going in. And for the most part I felt GA managed that expectation fairly well for most of its immersive programs - except Web Development.
In WDI I would see students who had no background in CS, design, etc. hoping to go from no experience to a full-stack developer position in 12 weeks. With a broad curriculum that iirc was starting with html/css fundamentals and working through Rails/Node/... maybe Angular? I felt it was giving them a base to start learning on their own, but not enough to immediately start working except for the very-top students (who likely had some sort of helpful background coming in).
I do believe GA is at its core a good organization that wants to see its students succeed (can't say enough good things about the career coaching / placement staff aka "outcomes"). But I can see how with a lot of competition in that space, the sell may have gotten stronger than what the product supports - and that's 100% on GA to address.
I don’t doubt that GA started with good intentions, but it seems clear that they got caught in the typical startup trap where they get ‘forced’ to prioritize investors over users as their target audience.
It wasn't outrageously hard stuff they were tripping up on, if you don't grasp closure or recursion or something that might be OK, but most couldn't code fizzbuzz. After having paid $10-15k in tuition and being trained for three months, I would at a minimum expect a student to be able to write a for loop.
Moreover, I blame them for completely weakening what "Data Scientist" is supposed to mean. It was never really clear in the first place, but always assumed to be "a programmer who knows about statistics or a statistician who knows about programming". GA distributes their certifications to people who are neither.
Corollary, you're better off coming to me with no diploma/certification than with a GA one.
What upset me were, as you said, those who wanted to believe the promise of instant-employability without ANY prior coding experience. But GA did not over-sell this. They make it very clear in the entrance interview that there are no promises and the employment process is on our shoulders.
I feel, perhaps, the industry oversells itself: claims of "everyone should learn to code", claims of developers "making millions on IPOs", and reading how big tech companies are making the headlines every day. My neighbor's are all driving nice, fancy cars. Why? Amazon developers. Who doesn't want to be a developer-turned-overnight-success with propaganda like that?
About a month into the course, to our despair, we discovered that there was almost zero conceptual training imparted (I am a data scientist by profession and can speak to this). The instructors seemed to be working there as a stopgap measure while looking for other jobs, and the students were there for standard credential-seeking. The course material meanwhile was focused on answering "How" but not "Why", which is far more important when approaching data science problems.
To fix the situation, we quickly pieced together a curriculum based on Tibshirani & Hastie's machine learning video lectures and picked a few Kaggle challenges for practice. My wife went through this material over a month, and came out with enough of a mental framework to be able to tackle book-length material such as ISLR.
Finally, there was almost zero career support provided apart from generic advice around how networking is important and how one should try to get outside their comfort zone.
Though all's well that ends well (my wife is working as a data scientist now at a seed-stage Valley startup), in retrospect we often talk about how we shouldn't have wasted money on this course. She didn't learn anything that she couldn't have picked up on her own, not even practical skills re: data analysis with Python/R, and her student cohort was quite disinterested in making connections.
Ce la vie.
 Introduction to Statistical Learning, http://www-bcf.usc.edu/~gareth/ISL/
The instructor was a "senior engineer" while his resume was only 2-3 years long, and he was an average at best instructor. The furniture was Ikea or Ikea-level. And perhaps most unforgivable to me, the (cheap) desks/benches did not have power outlets, even though laptops are a fundamental requirement. My last point may seem petty to some, but to me it indicated a distinct lack of consideration for what students actually need to succeed.
If I'm paying north of $100 per class hour, I expected considerably better.
The quality of the courses depends on the quality of the instructors -- and because GA tends to hire instructors as contractors without benefits, it's difficult for them to retain good instructors.
I also think the rigor and quality of the curriculum varies widely between campuses. NYC and DC do have fairly solid curriculums and high expectations for students. I get the impression that this is not the same for all campuses. (As an example: a campus on the west coast spends one hour total on SQL, while campuses on the east coast spend 9-10 hours of lecture time along with two homework assignments.)
I thought this was interesting. Anyone know more about the services General Assembly was offering to businesses? I guess there is more money and less risk associated with this market. Going after the public / the bootcamp model means you run into accreditation laws surrounding educational institutions and you have to do more work (e.g. tracking placement rates).
GA is a great org — they also did a lot of (job training) work with returning military vets as well. GA has gotten a lot of people into the tech world. Their courses aren’t all-encompassing, but their web dev courses are good enough to get junior devs off to a good start. I wish I had them when I got started! Their current Project Management courses look pretty solid as well. Highly recommended (but I have an admitted bias of course.)
1) outsource employee development
2) satisfy the need to hire lots of devs faster than recruiting could hire them
3) bring down the cost of hiring devs overall.
I left before seeing the results of the pilot, but I remember folks saying that it definitely opened up doors to those looking to switch careers, but that it's a slow process: it takes a lot of time and practice to get to even an entry level engineering position, especially when it's a part-time course.
Overall I like the idea of democratizing education and lowering the barrier people who want to get into the industry; plus, it definitely seems like a solution to the supply/demand problem for devs for companies, so I expect the trend of bootcamp/business partnerships to continue.
My codecademy course in data analysis seems to be a lot more work, but also a lot more useful.
In New York, at least, it was not accredited but it was...registered, I guess? with the state. I had to get a teachers' license.
The last round in 2015 was 70 million, that's a pretty good pay-off for 3 years, the one before that was 35 million in 2014, still a very good ROI.
Investors look for bigger multipliers when the risks are higher, a company like this is as start-up investments come reasonably low risk, it is essentially a middle ground between a consultancy business and a scalable proposition and as a result of that the returns are also in the middle.
Investors looking for bigger multipliers would have to be content with taking bigger risks as well.
The article actually mentioned a latest valuation iirc that said the last round investors had ~no positive return.
I agree with the evaluation of risk, though.
It's hard to know what they got back, but since the preference stack is noted I'd guess they made between 120% and 150%. There's no real way to know though - but we can tell they made their money back because the sale went though.
Also, A D round isn't looking for huge returns. They probably would have been outstandingly happy with 3x, so 1.2x is fine for them.
Some hints: the later investors likely had the best terms (they almost always do), they were most likely also in a position to block the deal if they did not make at least a certain ROI, they likely had liquidation preferences detailing exactly under what circumstances they would have to agree to a deal that was unfavorable to them and what the other shareholders would have to give up in compensation.
In my experience, programmers like solving problems and know that they are good at doing so. Most of my lecturers and classmates had spend many hours working out how to "fix" CS education -- it was the most immediate problem that they were exposed to every day. So it makes sense that many people threw themselves at the wall and not all of them stuck.
At HyperionDev , we're using a mentor-driven model to offer a fully online, part-time coding bootcamp where we give people significantly more 1:1 contact time with their mentor, and focus on code review as a primary method of instruction. We've found that we're able to teach people quite a lot in 6-months part-time at a significantly lower price point than some of the places mentioned in the parent.
The global bootcamp market might look a bit rocky, but until we get to a point where we don't have 10 dev jobs for every dev out there, bootcamps are going to have their place.
The problem with the bootcamp gold rush is and has always been the business model. You have a few major variables:
1. How much can a person pay you?
2. How many people can you fit in a classroom?
3. How long can the classes be while still maintaining a profit margin?
Based on those constraints bootcamps try to teach as much as possible. Within those constraints bootcamps found one model that worked and cloned it hundreds of times: A bunch of people in a classroom for 12 weeks paying $10-15k each. Some will try to chop off a couple weeks, some charge $18k instead of $10k, but it's fundamentally the same.
Every single bootcamp recognizes that this is not a great way to train engineers. 12 weeks simply isn't enough. But there's enough of a shortage or engineering talent that bootcamps could get away with it for a while.
Sure, in an ideal world you would actually train the students based on what they need to make hiring companies happy, but change any of the variables in the above equation and the whole business model falls apart. If you want to make the classes 2x the length you'd have to charge 2x the price, and can you get large numbers of people to pay $25-40k out of pocket? Not a chance.
We had to really start from first principles, and start with a new business model, which changes everything else.
The main thing we wanted was incentives aligned with our students; so we started from, "What if we didn't get paid unless students get a job?" That forced us to change everything else. I actually thought it might not be possible for a while.
I don't consider Lambda School to be a code bootcamp. We're six or seven months long full-time, which is 2.5x the length of a code bootcamp, plus a month of pre-course work. A few graduates have applied to Lambda School after having paid another school $15k and haven't even been able to pass our prep work.
We can be flexible on class size so long as we keep a good teacher:student ratios because we're entirely online (which, by the way, is actually a superior learning experience to offline). We can pick the best students regardless of financial ability because our pool of potential students isn't limited to people with 20k in their pockets that can move to San Francisco. We actually have _more_ support for students, because our instructors scale across classes when they're not teaching, and we can do some tricky financial stuff on the backend to make sure students are guaranteed a job before they pay.
All of that combined lets us open up a promise, which is what really matters for students at the end of the day - attend for free until you get hired.
I think we'll see massive consolidation in the bootcamp space in coming months, and it's already started (Dev Bootcamp, Iron Yard, Bloc, Viking Code School, and more have either been acquired or closed). That's very healthy, and frankly there are some sleazy bootcamps that only exist because of marketing. You could wipe the bottom 80% of bootcamps off the face of the earth and it would probably be a net positive for everyone. I think we'll get there in the next 3-4 years.
I assume "tricky financial stuff on the backend" means referral fees for placing students, and the like. Can you go into more detail on this?
I'm also curious how your online training works. Is it an inverted classroom model (read / watch stuff before class, discuss material and exercises in class?), or self-paced, or what? My experience is that online training doesn't work as well as face-to-face training, but this is in a very different context (training developers who already have jobs == they never do homework).
Everything we do is real-time, interactive, and absolutely not self-paced. There's supplemental material that's a flipped classroom model, but that's definitely not the main focus.
> The problem with the bootcamp gold rush is and has always been the business model. You have a few major variables:
> 1. How much can a person pay you?
> 2. How many people can you fit in a classroom?
>3. How long can the classes be while still maintaining a profit margin?
> Every single bootcamp recognizes that this is not a great way to train engineers. 12 weeks simply isn't enough.
> The main thing we wanted was incentives aligned with our students; so we started from, "What if we didn't get paid unless students get a job?" That forced us to change everything else. I actually thought it might not be possible for a while.
> I don't consider Lambda School to be a code bootcamp. We're six or seven months long full-time, which is 2.5x the length of a code bootcamp, plus a month of pre-course work
> All of that combined lets us open up a promise, which is what really matters for students at the end of the day -
attend for free until you get hired.
This isn't quite true. I work for Hack Reactor and we're doing well with this model. But every single staff member is also passionate about our students (including those at top), so that might explain why we work if others do not.
PS yall are awesome <3 love the material you keep putting out.
Austen is hella right about his thesis: bootcamps are not getting enough students across the finish line right now. They need higher admissions bars, more classroom hours, and/or other good ideas. Hack Reactor is (imo) one of the high-water marks of this line of thinking. Lambda and Holberton are others.
I think he's miscast the three variables though -- partly because he's got a bit of a dog in the race. (As I do -- "we're not a bootcamp" is something many bootcamp founders, yours truly included, have said.) Anyway here's another perspective on the matter:
#1 is mostly right -- bootcamps are limited in price, and this is a huge constraint. It's wrong in an important way: it should be "How much can I charge before a student picks another bootcamp". This problem gets easier, but doesn't go away, if a bootcamp uses a deferred-tuition model. Evidence for my formulation here: App Academy and Hack Reactor have the same kinds of student outcomes problems, because we're both facing the same demon, which is the low-cost bootcamp that prevents us from charging more and spending more on product quality.
Austen's #2 and #3 are both about specific ways to tweak the bootcamp's expense : quality ratio. This is an interesting topic that goes really deep, and I think Austen's reduction is (honestly) pretty heavy on the marketing content. If I were to boil it into two bullets it'd go like this.
#2 "Can you run a good program, online or offline, that isn't classroom-based." Here, by "classroom" I mean "15-40 students and teachers that know each other personally", and it can and does happen online. (Viking Code School, and Hack Reactor Remote, do online/classroom-based programs.) Classroom programs are very challenging to operate: they introduce discrete start dates, fill rate problems, etc. But it's hard to reproduce the quality of a classroom-based program in eg a mentor-driven or community-driven program. I think the answer here will be "kind of", and hybrid classroom-like programs will win.
#3 "Can you get students to pay $10k+ for an online program". Online can be better than offline -- Hack Reactor Remote is our top-performing campus right now. However, few people want to buy it, despite the fact that it doesn't involve moving to SF. Perhaps Lambda has figured something out here -- I don't know about their growth numbers. I hear Thinkful has. Generally, I think the bootcamp space will mostly become hybrid online/offline. I bet half of Lambda's students are in SF, they meet up in person already, and Lambda does or will facilitate/market this.
My #2 and #3 are also poor formulations of the fundamental challenge here -- the quality : cost ratio. For instance, my favorite recent innovation in the bootcamp space (evidenced by Holberton, and maybe Lambda?) is to offset tuition costs and increase grad expertise by bundling an internship and taking a bite of internship income. This doesn't pertain to my #2 or #3 (or Austen's) but it's a way to solve the top-line problem Austen mentioned.
Anyway, TLDR: another bootcamp founder agrees with Austen's thesis; quibbles on details in ways you might find interesting.
> #2 "Can you run a good program, online or offline, that isn't classroom-based."
In my opinion, every learning experience that requires teachers and/or a physical space is highly limited by its margins, and, as you said, will be much more sensitive to price-based competition.
If you want to increase the quality of the output, you need to increase the number of teacher-hours and real-estate-hours. Considering that labor and real estate are two of the most expensive resources you can think of, that is quite limiting.
Can we really think of a scenario where a high-quality learning experience is not limited by those two resources? I think Thinkful is doing a pretty good job. There is plenty of pedagogical evidence of the high impact that mentor-led education has (e.g. Bloom's 2 Sigma Problem). However, I think they are quite limited in a way that I consider crucial to solving the education problem:
They are very dependent on their mentor-led approach, both from a marketing and a financial point of view. That makes them very expensive and don't let them approach the problem/solution challenge from a more global point of view (i.e. only people in the US can, at scale, afford to pay $15k for a training like this, and their Income Share Agreement is only available to people in the US).
My main question here is this: What makes a mentor such an important element in the formula for student success? Is it the guiding, the technical knowledge, the accountability, the motivation?
My thesis here is that technical knowledge is not that important, but accountability, guiding and motivation are. At Microverse, we currently have students all around the world who are learning to code as part of distributed teams. They key here is that they spend almost 8 hours per day doing pair programming and holding each other accountable. We are "outsourcing" the task of holding students accountable to the students themselves.
However, there is also the motivation and the guiding aspect of the role of the mentor that students themselves can't take care of. In order to solve that, we are using quantitative/discrete input from the students that trigger the intervention of a more experienced mentor.
Also, one of my main hypothesis is that creating more content is not the key to adding value. There is already so much high-quality content available for free that only needs to be curated. And Thinkful (and almost every other player) is not understanding this part either.
Some students will think that they are paying for nothing if there are no teachers, no mentors, no physical space and no original content. However, if you flip the pricing in the way that Lambda School is doing by charging after the program, then the student perspective changes because she knows the only thing that matters is the outcome, and the payment is tied to that outcome.
We (Microverse) are the only training program that is currently offering an ISA available to anyone in the world. And there is a very simple reason why we can afford to do that: we don't have teachers, we don't create content, and we don't need to pay for real estate, all while making our remote experience accessible to everyone and while designing an experience from a motivation/accountability point of view through peer-to-peer work. All of this makes our margins way bigger, and that gives us much more room to take risks.
I could easily see this devolving into some kind of debt-based H1B system, where they drastically overprice a bootcamp education that they give for “free” to desperate jobseekers, and then use to lock them into underpaid roles at bad workplaces that can’t recruit technical talent without this kind of predatory leverage.
But NLP's accuracy isn't high enough for anything more than basic tasks.
It's just not good enough at understanding mistakes, or giving well calibrated probabilities of complex meanings given a context.
(I work in NLP professionally, and I've worked in the education space previously).
One thing to keep in mind, with something like this, employees' stock is almost certainly worth $0. There will likely be some sort of retention plan for some set of employees, but their stock in GA probably won't be worth anything. The founders will likely have some sort of bonus, and the investors will make some money (not home-run money, but money). I guess they deserve a congrats, but to me it's worth remembering that most likely employees, who viewed their stock options as a part of their compensation, will be left with nothing but a thank you.