What happens when the Board Of Directors begins to panic?
December 2, 2015 6:50 PM   Subscribe

I have been working with startups for most of the last 15 years, and one common pattern that I’ve seen is the startup that has a brilliant idea but terrible management. [...] I feel I can offer a real service by documenting my own experience and offering it up as a case study. I've spent the last 6 months working at Celolot, which falls into the "great idea/bad management" category. The idea is brilliant: Natural Language Processing as an interface to interact with big Customer Relationship Management tools such as SAP. The execution has been flawed. [...] All names have been changed except for mine.
via HN
posted by postcommunism (70 comments total) 17 users marked this as a favorite
 
This is a good narrative, although dude's insistence that he is the only sane person aboard this ship of fools is self-serving and not that believable.

I fully believe that everyone else has the personality flaws he attributes to them. But it is also obvious that he had no ability to move the needle on any of these vital management and strategic issues.

He's a senior member of a, what, 4-person team? And he can't successfully communicate any of this stuff to his teammates? Why does he allow himself to be such a passive victim?

Ultimately this is a very thorough documentation of how everything is someone else's fault. And that's a fundamentally suspicious genre. Would you want to hire someone who wrote one of these?
posted by grobstein at 7:30 PM on December 2, 2015 [34 favorites]


The idea is brilliant: Natural Language Processing as an interface to interact with big Customer Relationship Management tools such as SAP

There is no amount of money that could get me out of bed every morning to work on this "brilliant" idea. We've become a nation of Initechs
posted by any major dude at 7:38 PM on December 2, 2015 [43 favorites]


Given that SAP is a wall of fail to start with since there's always custom stuff and it never gets finished because the beancounters think SAP is turnkey (seriously every SAP add-on dev I've ever talked to has said this), the idea of pushing the UI to SAP through natural language is the kind of thing that nobody in their right mind would expect to work.
posted by Bringer Tom at 7:39 PM on December 2, 2015 [25 favorites]


I want to know who the kid Pranab is related to on the board that explains why he is untouchable.

God, I am having eye twitches and deja vu just reading that story. Management promises something impossible in the teeth of the developer estimates and then you have to nearly kill yourself to get it done while they scream at you for not being fast enough is the story of development.
posted by winna at 7:40 PM on December 2, 2015 [9 favorites]


That idea is deeply stupid and the investors deserve everything they get.
posted by abulafa at 7:53 PM on December 2, 2015 [6 favorites]


The author is a friend of a friend so I read this right after he posted it. At the beginning I thought he seemed sympathetic but by the end I shared my friend's reaction: he should have waited until he was less upset to write this piece, and possibly by the time he was less upset he would not have written it at all.

Still, kind of a fun read from a gossipy perspective. And it made me feel better about the chances of my husband's startup succeeding... yes, most startups fail, but most startups look a lot more like this story than my husband's company.

In any case... their product idea is both fairly useless unless done really well, and way harder to do really well than any of them seemed to think it was.
posted by town of cats at 7:59 PM on December 2, 2015 [7 favorites]


Lawrence is naive. There is a reason Siri isn't better, it's a really really really hard problem that 4.5 people and tight deadlines are not going to solve.
posted by sety at 8:03 PM on December 2, 2015 [20 favorites]


Hinton (our 22 year-old CEO) ...

Oh fuck me. I've known a lot of talented 22 year-olds but there's no way that I'm signing on to a company run by one.
posted by octothorpe at 8:05 PM on December 2, 2015 [11 favorites]


What struck me is that when Lawrence comments on how the newsletter that Pranab is asked to write is poorly-written, like his code, is that if I apply that same lesson to Lawrence's writing here, I wouldn't want him writing code, either.
posted by jimw at 8:11 PM on December 2, 2015 [5 favorites]


grobstein: "dude's insistence that he is the only sane person aboard this ship of fools is self-serving and not that believable."

I didn't really get that as the message... "Shinzo", "Gregory", and the consultant are all spoken of as reasonable people. Even "Pranab" isn't irrational, he's just underqualified for his job and somewhat of a slacker.

And the board (with the exception of "Milburn") isn't really vilified either: they were simply given bad information. Now, one might argue that the board of a company has a responsibility to demand verifiable information, or at least corroborate things from multiple independent sources. To that argument, I reply: ¯\_(ツ)_/¯

I have no doubt that Lawrence made mistakes. Small ones, of the kind that inspire commit messages like "I am an idiot". And large strategic ones, like getting drawn into a debate about an implementation detail when a broad change to the implementation as a whole is what needs discussion. I admit that this blog post would be well-served by such anecdotes. </🍔>

To be frank, this seemed like a hedged, introspective account of events I have witnessed in too many situations—both professional (liaisons who only advocate for the "powerful" side), and personal (rapid tone changes between cold reason, spitting rage, and wounded puppy). The details are probably coated with some of that ex post facto academic analysis, but the shape is clear and familiar.

I believe this story.
posted by Riki tiki at 8:17 PM on December 2, 2015 [8 favorites]


That was kind of vicariously cathartic, that final insane conversation was clearly playing on repeat in his head.

It does sound like an impossible project, but there's a lot of them about.
posted by lucidium at 8:21 PM on December 2, 2015 [3 favorites]


Oh fuck me. I've known a lot of talented 22 year-olds but there's no way that I'm signing on to a company run by one.

Not necessarily fair, there have been a lot of really successful companies with very young founding partners, and I have worked for some and so has my husband.

From my POV, with a history on the admin side of startups and a secondhand view to the more technical stuff, I can only relate to this so far, but it does resonate. It also makes me really appreciate the startups I've been involved with that have been much more focused on putting the right people in the right places, and actually, when he talked about needing a good project manager to get things in place, that's where being involved in even an ultimately unsuccessful venture has shining moments. I think "project manager" can sound like a really meaningless role but the people who are actually doing it well are helping everyone around them be better at their jobs going forward.
posted by padraigin at 8:21 PM on December 2, 2015 [3 favorites]


Lawrence is naive. There is a reason Siri isn't better, it's a really really really hard problem that 4.5 people and tight deadlines are not going to solve.

Yeah, the part where he's describing why their project is way, way harder to do than Siri surprised me. You knew that and signed up anyway? What made you think you could pull it off? And it especially kills me that Pranab was responsible for all the NLP stuff. You thought a kid with no experience or expertise could do this, by snapping together publicly available libraries?

Will skilled programmers really take jobs with such obviously flimsy prospects? I guess they had funding, so at least there was a salary. But what were the ex ante odds that this "multi-billion dollar" business would so much as release a product?
posted by grobstein at 8:22 PM on December 2, 2015 [6 favorites]


There is a reason Siri isn't better, it's a really really really hard problem

Well, they probably need to simplify a little. Design some basic rules that restrict the input language to a smaller subset of natural english. Make the grammar suitable for dealing with systems beyond just SAP, so it's a common language. A common language for business-oriented logic. They could call it COMBOL.
posted by sfenders at 8:24 PM on December 2, 2015 [22 favorites]


There is no amount of money that could get me out of bed every morning to work on this "brilliant" idea. We've become a nation of Initechs

I'd be happy to work on NLP for its own sake I don't know what the hell the rest of that idea is about.
posted by atoxyl at 9:03 PM on December 2, 2015


Milburn was right, though. How the hell are you supposed to write a coherent program without a spec and a set of requirements derived from that spec? I mean, I know it's fun to be a cowboy coder out on the range, but among other things this seems like a case study in awful engineering practices.
posted by johnnydummkopf at 9:06 PM on December 2, 2015 [5 favorites]


So the dude anonymized everything but himself, in 2015, on his own blog. Which links to his real LinkedIn. Which appears to have dropped the last six months off his profile, everywhere but the bit that says "Lead Programmer at Rollio Force."

Honestly, Crunchbase shows them as fucked anyways. They have a grand total of half a million dollars in the bank. The burn rate in NYC is not going to sustain the company for as long as Lawrence wants to finish the product without another round of financing, and that means hacky tech demos, and possibly beta users.
posted by pwnguin at 9:11 PM on December 2, 2015 [5 favorites]


I really enjoyed reading this and thought the narrator came across as a pretty decent person to work with. He comes across as being willing to dig in and do the hard work, generally deferential to the power structure but also willing to call things like he sees them, basically realistic about time without wanting to spend a lot of his own time on management, and able to stay calm and pick his battles even while being yelled at.

I do wonder about his motivations for writing this. Is airing dirty laundry like this de rigeur in the startup world? And if not, what would cause him to override that -- preemptively defending his reputation? Or, given the ending, is he trying to sell the idea and his experience advancing it to another set of investors? Is it written for the Board's eyes -- would he like them to get rid of Hinton and Milburn and bring him back on the project?
posted by salvia at 9:12 PM on December 2, 2015 [1 favorite]


pwnguin: "So the dude anonymized everything but himself, in 2015, on his own blog. Which links to his real LinkedIn. Which appears to have dropped the last six months off his profile, everywhere but the bit that says "Lead Programmer at Rollio Force." "

Which, if you poke around a bit, has a grand total of three other software engineers: an Asian guy, an Indian guy, and a white guy (listed as a "Software Engineering Consultant", so potentially part-time). I guess the pseudonymization isn't really so much to hide their real identities but rather to protect their Googlability.
posted by mhum at 9:38 PM on December 2, 2015 [3 favorites]


Ugh, reading this was like being back at my last job.

Here's a chatfilter question: has anyone here in a technical field ever worked well with a boss who was great at sales but wasn't quite up to speed on the tech stuff?
posted by infinitewindow at 9:58 PM on December 2, 2015 [1 favorite]


johnnydummkopf: "How the hell are you supposed to write a coherent program without a spec and a set of requirements derived from that spec?"

Many of the interpretations of "agile" I've seen behave exactly like this. And that's because agile puts a nice spin on "get to work right now."

It's supposed to also come with honest retrospectives, periodic reassessments of priorities and planning (with buy-in from both stakeholders and implementers), and genuine investment in architecture and design during development. However, like most project management methods, project managers are the ones who deal with it most directly, and they're primarily accountable to senior management.

But senior management—especially in the instant-gratification clusterfuck of startup culture—doesn't care about architecture, design, team buy-in, or agile. They have to appease their gods (though I guess we should commend their humility in only describing themselves as "angel" investors) and that means shipping yesterday.

This is catastrophic when they don't understand the challenges and don't trust the people on the ground. It leads them to circumvent the team, willfully avoid unpleasant problems, and act shocked and play the blame game when things inevitably unravel.

That's why the early strategic interactions are so important for tech folks to get right. Honestly, this is one of the things a lot of developers really shirk: we just want to code and don't personally care about agile vs. waterfall, estimates and story points, etc. But as part of the team, we have a professional obligation to be invested in the boring big-picture details as well. Because we don't, we have partially earned our reputation as people who waste a lot of time going down rabbit holes.

That said, the whole point of senior management being paid the Big Bucks is that they're supposed to be responsible for the whole. That means they should be the ones recognizing and addressing these systemic issues. They should be encouraging developers to raise their concerns early, but instead usually end up (directly or indirectly) punishing them for doing so.

Had the developers been given time at the beginning of the project to do due diligence, they could've drafted the initial spec and requirements and made sure everyone understood it. But with the heat on right from the start, they're seemingly left with no choice but to put those things on the permanent backlog.
posted by Riki tiki at 10:01 PM on December 2, 2015 [14 favorites]


Heh, the visual metaphor for their system in the video is a bunch of unconnected cogs turning at different speeds.
posted by lucidium at 10:02 PM on December 2, 2015 [11 favorites]


I feel for the kid, though every story of development failure seems like a replay of The Mythical Man-Month, which is now 40 years old. But then...

The idea is brilliant: Natural Language Processing as an interface to interact with big Customer Relationship Management tools such as SAP.

Oy. He's basically asking for AI-- old-school AI of the Winograd toy-world type... and he's surprised that it couldn't be done in six months.

And suppose it had worked? Suppose you had an app where you actually had to type "I want to edit the info about my contact Jenny Hei"? What a horrible idea. Or maybe it is brilliant, in that it would open the door for yet another app that would condense that down to EDIT CONTACT JENNY HEI, and maybe another app on top of that would replace it with a Contacts button.
posted by zompist at 10:04 PM on December 2, 2015 [14 favorites]


but very little feels justifiable in the face of a crisis, except writing code as fast as possible.

This is such a dumb statement it could only be written by someone who knows how to code but nothing else.

The superstar consultant sounded like the only fullwit in the whole story. NLP is easy! SAP just needs a new frontend! Open source libraries are good to use as engines for commercial software! The fact that these clowns got funding speaks to how much free money is sloshing around looking for a place to park it.
posted by benzenedream at 11:16 PM on December 2, 2015 [5 favorites]


Development blames Management: deadlines created according to business needs are unrealistic.

Management blames Development: agreed deadlines are not being hit and the junior guy's the scapegoat.

Reality is, a team of 4 developers is not going to exceed in 6 months what Apple did with Siri. The problem isn't management, it's that the problem is Hard. The author can blame managment for not seeing that, but from my perspective it looks like he's been drinking the kool-aid too.
posted by cotterpin at 11:38 PM on December 2, 2015


So, after getting screamed at by Milburn for being unable to take responsibility and for blaming everyone else, his response is to... write a blog post that takes no responsibility and blames everyone else? Look, I have no doubt that he's being asked to do the near-impossible, and that he's working under some pretty awful people, but I really suspect we're not hearing the full story here.
posted by Guernsey Halleck at 11:44 PM on December 2, 2015 [3 favorites]


Leaving aside the degree to which the whole narrative is self-serving, even taken at face value; it was hard for me, by the end, not to feel like Lawrence was enabling the whole train wreck of clowns. The project would have been ludicrously difficult under ideal startup conditions where everything is running well, and Lawrence was plainly aware early on that the setup was subpar even by startup standards. Tolerating Milburn's behaviour after all of that, for weeks on end, puts it squarely into the territory of the delusional or the masochistic.
posted by fatbird at 11:49 PM on December 2, 2015 [2 favorites]


Everyone involved seems hopelessly naive, including (if not especially) the narrator. They all seem to have thrown common sense out the window in order to chase the idea of a ludicrous amount of money. I'm reminded of Warren Buffet's famous advice to never invest in a company one can't understand. The author doesn't seem to understand why the venture really failed and has done himself a disservice as far as future employment - I would be extremely hesitant to hire someone that posted something like this.
posted by Candleman at 12:05 AM on December 3, 2015 [2 favorites]


Yeah I was hired to do some documentation for a startup that was trying to make a CRM with 2 "business guys" and one programmer, and then me writing the documentation. They were astonished to find that you couldn't just slap together a Salesforce knockoff in six months and ship it. Programmer told me to keep my head down and just keep collecting the money, which I did. And that's not even my "best" startup experience, which is the time they laid off all the people who actually did work (programmers, CS, project managers, etc.), kept all the C-levels and management, and were astonished that all the customers bailed on them.
posted by Ghostride The Whip at 12:09 AM on December 3, 2015 [1 favorite]


"Salespeople hate software. They are good dealing with people, but they hate buttons and forms and all the other junk that is part of dealing with software. A pure chat interface eliminated all of that debris that salespeople dislike."

Funny how the problem they were trying to solve with their product so closely mirrors the cause of their own problems. Salespeople who can get away with just handwaving their wishes, while leaving the messy details of implementation to somebody or something else.
posted by spudsilo at 12:15 AM on December 3, 2015 [10 favorites]


Here's a chatfilter question: has anyone here in a technical field ever worked well with a boss who was great at sales but wasn't quite up to speed on the tech stuff?
That depends on whether there's a significant level of trust and mutual respect.

In other words "no".
posted by fullerine at 12:27 AM on December 3, 2015 [1 favorite]


I'm left scratching my head at some of the tools they used. For example, what was the benefit of going with Clojure? It sounds like the NLP had to leverage Java, and you'll find the most developer support on Android with Java. So right away two key pieces of the project are a best fit for Java... what does Clojure bring to the table?
posted by sbutler at 12:27 AM on December 3, 2015 [1 favorite]


I'm left scratching my head at some of the tools they used. For example, what was the benefit of going with Clojure? It sounds like the NLP had to leverage Java, and you'll find the most developer support on Android with Java. So right away two key pieces of the project are a best fit for Java... what does Clojure bring to the table?

Clojure is on the JVM so it should interface pretty well with the Java components. I'm not sure that particular choice of language helped them out though considering the apparent shallowness of dev experience.
posted by atoxyl at 12:37 AM on December 3, 2015 [3 favorites]


This is fun and educational to read for its insights into tech culture's failure modes (building a cutting edge NLP app without an NLP expert! Trying to outdo Siri in a few months of dev time! Legerdemain and crunch time for demos!) ...but there are way too many places where the amount of interpersonal detail and blame-gaming makes me deeply uncomfortable. I feel like I'm being made a part of someone's years-long argument with their spouse. I would feel a lot better if those parts were edited out. They do no one any good.
posted by daveliepmann at 12:45 AM on December 3, 2015 [4 favorites]


The whole thing just gives me anxiety. I think most programmers have experienced a non-technical boss blithely over-promising to clients and also probably some times where they themselves severely underestimated the time required to complete a conponent. And that's all pretty bad. But they're just so clearly in over their heads from the beginning here. Special note of the "pivot" from implausible project to impossible project.
posted by atoxyl at 12:59 AM on December 3, 2015 [1 favorite]


It's a good thing SAP doesn't own a company that does NLP and finite state machine work.
posted by benzenedream at 1:31 AM on December 3, 2015 [5 favorites]


Perhaps Pranab's unfireability comes from a simple misunderstanding -

"But it says on his resumé that he's an NLP Master!"

"Did you explore exactly what he means by NLP?"

"Oh."

"Surely the fedora should have given it away."

And I'm reading it as something semi-fictionalised. Apart from anything else, The Idiot - under-qualified and slacking - is a role that someone on a team is required to adopt.

That said, the conversation with Milburn was quite eye-opening for me, in my relative naivety. I've had similar conversations with managers, and I thought they were ... insane is probably a bit strong, but you know what I mean. I didn't realise it was a strategy. I still can't see what the strategy is supposed to achieve. Perhaps it's employed by people who want to maintain total control over something they don't actually understand.

Another thing that struck me is the number of issues that become confrontations that form to allow only two responses - either submission or defiance. This is something that I've been noticing in life as well, and I've been trying to work out what a third response might be. Tricky, since they seem to arise with the Milburns of the world, for whom it's a win-win situation (either you surrender or they defeat you).

Also, the character Milburn seems like modern capitalism in micro: it seems like his aim is to create a thing, but in fact what he's done is engineer a situation that he can go into and extract as much from as possible leaving the underlings in the ruins to clean up the mess and carry the blame. I'm fond of saying that the word "Entrepreneur" comes from the French words entre, which means "between" and the verb prener which means "to take". In essence, an entrepreneur is someone who gets in there and takes as much as they can. I assume that there was a time that capitalism was about more than simple rent-seeking and rent extraction, but even if it was, that's no longer the case.
posted by Grangousier at 2:13 AM on December 3, 2015 [6 favorites]


I don't know why the fake name of each character was required to remain super ethnically identifiable. Is his mental model of the other programmers really so strongly focused on them being "the Asian guy" or "the Indian guy"?
posted by the agents of KAOS at 2:28 AM on December 3, 2015 [1 favorite]


I can't believe there was no final reveal about Pranab. You can't repeatedly say "I thought it was odd that they kept covering for him" and then just leave that gun hanging over the fireplace at the end. This is possibly the most outrageous part of the whole story.
posted by No-sword at 3:26 AM on December 3, 2015 [7 favorites]


"The idea is brilliant: Natural Language Processing as an interface to interact with big Customer Relationship Management tools such as SAP."

The idea is brilliant? No it isn't. It's bolting more trendy crap like 'ooo Facebook integration' onto ERP, and ERP is essentially a solved problem.
posted by GallonOfAlan at 3:44 AM on December 3, 2015 [2 favorites]


Another thing that struck me is the number of issues that become confrontations that form to allow only two responses - either submission or defiance. This is something that I've been noticing in life as well, and I've been trying to work out what a third response might be. Tricky, since they seem to arise with the Milburns of the world, for whom it's a win-win situation (either you surrender or they defeat you).
The third response you're looking for is to define and enforce boundaries. You make these confrontations about these boundaries, not whatever issue the Milburn has picked out. You give him the choice: either he lets you do the job you were hired to do, he doesn't trust you to do it and he fires you, or he continues the bullshit and you quit.
posted by cotterpin at 4:12 AM on December 3, 2015 [8 favorites]


I have worked for (actually, was Employee #1 at) a start-up, where the two owners were basically tech and marketing/sales. Both were brilliant, both trusted and respected each other, and both knew their own strengths and weaknesses in the other person's speciality. We had many ups and downs, went down a few blind alleys and got a lot of things wrong, but it was overall a very positive experience. We got product out, it did what customers expected and they really liked what they got.

The company grew to about ten people and our Big Product was in many ways a real success - when things fell apart, it was for external reasons largely beyond our control. But as a small start-up tech company producing real things in hardware and software that were in every way comparable with what a big corporation would have done in the space, were genuinely innovative and which I can still point to with pride, it worked very well indeed.

It may be a rare thing, but yes, you can do good things with good people. We had exceptional people at the top (the tech guy especially, who combined top-notch hardware and software chops with absolutely no ego and an exceptionally clear-headed management style) and I'd still go and work for them again at the drop of a hat.
posted by Devonian at 6:07 AM on December 3, 2015 [3 favorites]


This brought me right back to a startup I was a pm at a few years ago, where we had a really exciting concept to work with and the team to pull it off. But just like in the article, the sales people at the top were just as impatient. Oh the pivots! And the manipulative conversations were virtually identical to the ones in the article.

I don't miss the tech startup industry at all where the same guys get more and more vc funding for their latest idea, the vc's get to feel cool, and business meetings are treated as viable avenues for dating (unbeknownst to me until that awkward moment where they want to talk about my beautiful eyes instead of my app).

Incidentally, my direct boss at the exciting but toxic startup was a sales guy, not a tech, but he was the best manager I've ever worked for. Unlike his bosses, he understood (1) how to trust his team, and (2) how to create a healthy company culture. Even his sales strategy was more focused on customer experience not "sales". I wish he had been at the top instead, because we might have managed to finish instead of crash and burn.
posted by A hidden well at 6:17 AM on December 3, 2015 [2 favorites]


Coincidentally, I just came across an article on the head of Squarespace - demonstrating a very different kind of leadership and how to create a healthy workplace
http://www.entrepreneur.com/article/247417
posted by A hidden well at 6:49 AM on December 3, 2015 [1 favorite]


I was nominally the tech “lead” but I started with the assumption that we were each excellent at our own tasks, and so each of us could be trusted to build out the part of the system which we had been hired to build.

Aside from the merits of whether this was possible or not, this is the sentence that said to me that this project was irretrievably doomed from conception. There never was a true lead on a project. This author may have the best technical chops ever, but he has no clue how to deliver a project bigger than one person.

He had a CEO and board that had no idea what they were doing either, but he had no business pretending to run a team. He doesn't concern himself with ensuring the work of others was on track or even competent. His main arguments when it falls apart consist of "it wasn't me!".

I've worked with people like this author. They're often great technical specialists. They work great in a transactional solo mode. However, if they're unwilling to step up and take the reins for the larger project, as the author was, they're not a "lead" in any way shape or form. There are many strategies and boundaries he could have insisted happen with Hinton, disciplining or firing their game-playing intern, being present at demos, etc..., but he avoided them. He was more interested in justifying his own correctness than making the project work. He's possibly a great employee in a limited role with no responsibility, but his passivity and unwillingness to own the project make him temperamentally unsuited to do much more.
posted by bonehead at 7:17 AM on December 3, 2015 [16 favorites]


The idea is brilliant: Natural Language Processing as an interface to interact with big Customer Relationship Management tools such as SAP.

...

Instead of the user clicking a button to edit old Contacts, they would type, " want to edit the info about my contact Jenny Hei," and we would then find the info about Jenny Hei and offer it to the user so they could edit it. That was the new plan. This was a brilliant idea. Salespeople hate software. They are good dealing with people, but they hate buttons and forms and all the other junk that is part of dealing with software.


So... your sales guys hate using software because their intentions get muddled by the software misinterpreting or actively impeding what they want to do, so your solution is to recreate the UI from Quest For Glory?

"Brilliant." You keep using that word. I do not think it means what you think it means.
posted by Mayor West at 7:26 AM on December 3, 2015 [2 favorites]


I don't get why the NLP engine has to be so flexible. Why not just require the user to do more meta-management of the conversation, using keywords and triggers? We do this in actual conversation with other humans -- refreshing our interlocutors memory -- so it wouldn't trigger the aversion to software in salespeople that Ceolot/Rollio was trying to avoid.


> I want to update the Hilton deal from yesterday

{Do you mean the deal with Hilton West BFE on Whatsis Ave, created 00/00/00, contact: A. Buyer?}

Yes

{What do you want to update?}

Choices?

{You can update

Foo
Bar
Baz
Quux

Here is the required information currently missing:
Baz
}

Update Baz

{What is the new Baz for the Hilton deal from yesterday?}

The new baz is a gajillion.

{I've updated the Baz for the Hilton deal from yesterday to a gajillion. This deal may now be submitted.}

Submit the deal.

{Please say 'confirm' to verify you want to submit the Hilton deal from yesterday.}

Confirm.

{I've submitted the deal. Your new sales total for Q4 is $______.}


etc.
posted by snuffleupagus at 7:30 AM on December 3, 2015


Can it do "Earl Grey, hot"?
posted by bonehead at 7:32 AM on December 3, 2015 [1 favorite]


Can it do "Earl Grey, hot"?
You get a cup of no tea.
posted by k5.user at 7:37 AM on December 3, 2015 [3 favorites]


Other people have said it, but the idea sounded flawed to me from the jump. Replacing a UI with NLP? That works for Siri because you're effectively replacing a single text field and button, and it's backed up by Apple's money and expertise. Sure, it would be a great thing to have, if you can pull it off - but I'm trying and failing to think of any other piece of software that does the same thing. Although I think Lawrence's retelling of events is pretty easy to sympathize with, I would have seriously considered bailing at the moment of the Big Pivot, especially with the decision (and timetables!) being made without any input from engineering. That's a Really Bad Sign.
posted by Edgewise at 7:51 AM on December 3, 2015


I don't get why the NLP engine has to be so flexible.

I might have misread Lawrence's presentation of the issue, but the basic technical difficulty had to do with tracking conversational state at all, and the kind of technical challenges that raises in an environment with non-deterministic networking behaviour and a central server doing the NLP. However much meta-management is necessary is a problem downstream of that core issue.

And basically, their sales proposition was going to be that you don't need a lot of meta-management at all. And I can imagine a chain of technologies that, if it works out, yields pretty clear conversational state. In my limited exposure to NLP, once you had a parsed sentence, you were in a pretty strong position to apply context and draw inferences. It was just difficult to get to a clearly parsed sentence that wasn't contextually ambiguous to start with.
posted by fatbird at 8:53 AM on December 3, 2015


I can't believe there was no final reveal about Pranab. You can't repeatedly say "I thought it was odd that they kept covering for him" and then just leave that gun hanging over the fireplace at the end. This is possibly the most outrageous part of the whole story.

My guess is either that the VC that backed this with Milburn wanted him there. Or Milburn thinks his time with Goldman-Sachs will be valuable for another round of financing. Or maybe their office space at the NYU incubator requires some level of student involvement?
posted by pwnguin at 9:00 AM on December 3, 2015 [2 favorites]


I feel bad for the team. The situation was not optimal for anyone. What I find fascinating was the author's dissection of Milburn's communication techniques because it is how sales reps work.

In a previous life, I dealt with software sales and renewal reps. Here is the thing about sales reps: it is the company's interest to know the complete flow of the sale NOT the sales rep's. The sales rep does not want the company to have their complete Rolodex, or monitor all of their moves. So you have to think about, very clearly, how to incentivize the sales rep to showing all their cards and not revert to post it notes and off record calls, lunches, meetings, etc. A sales rep's goal is to make money for themselves, by any means necessary. Unless that software is very necessary to that process it is a hard sell for reliable compliance. So, the author's sales rep theory of mind seemed a bit limited. Of course, sales reps hate bad UI, doesn't everyone?

Another thing I noticed, there were movie level tropes of who is at a start-up. Young CEO still in school, check; the young developers: East Asian, South Asian, the white guy tech lead; brilliant software; crunch time extreme; older guy running the show and money being burned. Are boards and VC money enthralled with the veneer of what a start up should look like? The placement agent?I guess, I am trying to wrap my head around this from the perception angle.

Anyway, I thought the article was an interesting glimpse into the present start-up culture. I hope he does not just curl his own career.
posted by jadepearl at 9:11 AM on December 3, 2015 [2 favorites]


I understand how writing this might have been cathartic for the author. But no one comes out of this looking good, author included.
posted by danhon at 9:31 AM on December 3, 2015 [4 favorites]


I do not work in software development in any way, I worked for an engineering consulting company in oil and gas, and man is this very familiar. Organizational dysfunction really transcends industries.

This was a project management train wreck from the beginning, obvious in hindsight and a pattern that I have seen repeated over and over. On a lot of smaller engineering projects -- ones that are small enough that it is hard to justify having a full time project manager -- it is common to assign pm duties to the technical lead, and in all the projects I can think of off the top of my head things went pretty much like Lawrence's story: The technical lead assumes everyone else knows what to do and grossly underestimates how much actual project management will be needed, gets bogged down in technical work, and everything goes off the rails. It has happened to me too. I think part of it is that when you become an expert in something you forget what non-experts don't know, and you over-estimate the knowledge base of your junior co-workers. Also it is very easy to assume that what is obvious to you is obvious to everyone (e.g. what the priorities in the project are, what the next steps are), but this is rarely the case.

This throw-away line bugged me "If Celolot had been officially committed to running a mentorship program for junior programmers, then of course I would have understood that this was a responsibility of mine. However, we were not running a mentorship program." This is an odd position for a lead to take, whether you have a formal mentorship program or not mentoring junior personnel is always part of the job of being a good lead. Sure this guy could have been just exceptionally terrible, which I think was the point he was trying to make, but this blanket statement that it wasn't his job to mentor his subordinates struck me as plainly wrong.

I have worked with many people like Milburn, and that emotional roller coaster from almost crying to screaming and yelling about nothing typically reaches its crescendo when the money is running out. My experience was with oil and gas companies, so on the one side is emotional abuse by senior management and on the other, well, we can all read the newspaper and see that oil prices have dropped dramatically and the industry in dire financial straits. All the emotional breakdowns ever accomplished was poisoning whatever good will was left and convincing the best employees to flee before the inevitable lay-offs. If there is a hope for rescuing the company, the Milburns of the world destroy it. (In my case, as a consultant, I got paid either way so I had some professional distance while watching the slow dissolution of the clients company. Thankfully my company didn't have any Milburns, unfortunately they didn't have any work either, and I was also laid off.)
posted by selenized at 9:59 AM on December 3, 2015 [6 favorites]


I'm a Salesforce Consultant and the idea is not brilliant, it's laughable.

Salespeople loathe updating their CRM, that's fact. This app is easily 10 years away from being rudimentarily usable for an ERP/CRM, and it would only be those that are not doing any customization. So that market would be 0%.

SAP is so complex, so customized and so messed up....I can't believe anyone would think that this would be go-to-market ready within a decade, let alone 12 months.

Our narrator has done an excellent job in showing us exactly the guy who thinks he knows what he's asking for, but in fact, hasn't a clue. One of my last gigs was with a manager who insisted that we consider the "Service" module for their instance of Salesforce. I was confused, and naturally so, as they used Cases with Web-to-Case.

Me: You're already using Service. That's the part of the software that deals with Cases.

Mgr: No one ever said that. Everyone tells us we should upgrade to Service.

Me: You have service, perhaps they mean Service Console, but you're already using a soft-phone vendor's console and your Service folks love it.

Mgr: We should have Salesforce come by and do a demo.

It's no big wonder that job only lasted 3 months. And, FWIW, I felt that exact amount of frustration with that group of managers. Complete with anxiety attacks, and questioning my sanity.

Yeah, this should probably have a trigger warning.
posted by Ruthless Bunny at 10:26 AM on December 3, 2015 [6 favorites]


the agents of KAOS: "I don't know why the fake name of each character was required to remain super ethnically identifiable. Is his mental model of the other programmers really so strongly focused on them being "the Asian guy" or "the Indian guy"?"

I feel like this is an unwinnable situation. If he had called "the Indian guy" Steve instead, you could accuse him of trying to portray everyone as white guys.
posted by Chrysostom at 10:40 AM on December 3, 2015 [2 favorites]


This is an odd position for a lead to take, whether you have a formal mentorship program or not mentoring junior personnel is always part of the job of being a good lead.

That jumped out at me too. It's a symptom of how badly suited he was to the role. It's a common attitude for an employee without decision-making responsibility (and is perfectly fine there), but that sort of hidebound thinking in a PM is fatal.
posted by bonehead at 10:43 AM on December 3, 2015 [1 favorite]


Chrysostom: "I feel like this is an unwinnable situation."

Well, he could have used the 19th century, "based on a true story", roman-a-clef style of referring to people by initials, e.g.: Our CEO, Mr. A_____ and his wealthy benefactor Mr. V______, etc...
posted by mhum at 10:51 AM on December 3, 2015 [3 favorites]


I think what happened with the writer is that they hadn't ever managed a project before and sold themselves on the basis solely of their technical expertise.

Everyone thinks project management is easy until they actually do it.

I'd've also have liked to see the author break down his personal opportunities to improve: it was an unwinnable situation but some of his choices made it worse.
posted by winna at 10:54 AM on December 3, 2015 [4 favorites]


I feel like this is an unwinnable situation. If he had called "the Indian guy" Steve instead, you could accuse him of trying to portray everyone as white guys.

I would probably have portrayed an ethnically diverse team but shuffled the names around completely. Did he not do that?
posted by atoxyl at 11:21 AM on December 3, 2015


I'm still confused on how it took 5 months to set up a wiki to document the data structures (and hopefully the API defs of each service!).
posted by ryoshu at 11:33 AM on December 3, 2015


It didn't: they're using github, which has a wiki built in for a project, and they were updating their own code there as they went, but Lawrence had to keep spanking them to keep updating it.
posted by fatbird at 11:39 AM on December 3, 2015


scotty_hello_computer.gif
posted by Damienmce at 12:46 PM on December 3, 2015 [2 favorites]


I do not work in software development in any way, I worked for an engineering consulting company in oil and gas, and man is this very familiar. Organizational dysfunction really transcends industries.

I work in public policy - a long way from anything engineering-related - and a lot of it rings true to me, especially this:

On a lot of smaller engineering projects -- ones that are small enough that it is hard to justify having a full time project manager -- it is common to assign pm duties to the technical lead, and in all the projects I can think of off the top of my head things went pretty much like Lawrence's story: The technical lead assumes everyone else knows what to do and grossly underestimates how much actual project management will be needed, gets bogged down in technical work, and everything goes off the rails.
posted by une_heure_pleine at 1:26 PM on December 3, 2015 [1 favorite]


I would probably have portrayed an ethnically diverse team but shuffled the names around completely. Did he not do that?

Nope, you can see the other employees on LinkedIn and they appear to match his ethnicity/responsiblities breakdown. That's what made me wtf.
posted by the agents of KAOS at 1:30 PM on December 3, 2015


If I had been writing the article I think I would have left ethnicity out of it entirely. It adds nothing to the story or the points he was making.
posted by sbutler at 2:13 PM on December 3, 2015


I too have had a "we have decided that this untested thing is now all of the things" experience.

It did not work.
posted by ckape at 2:16 PM on December 3, 2015


Did I ever tell you guys the story of the time I got sent, as the most junior guy in the firm, to mentor a bunch of old-enough-to-be-my-father COBOL programmers in writing VB to screen scrape an AS/400 supply chain management application? That went only a little bit better than this story.
posted by ob1quixote at 4:03 PM on December 3, 2015 [2 favorites]


So this is sociopath engineer vs. sociopath salesman? Kind of an Alien vs Predator, but without the nuance?
posted by elwoodwiles at 8:50 PM on December 3, 2015 [2 favorites]


« Older Horror Victorianorum   |   Lagavulin™ Newer »


This thread has been archived and is closed to new comments