I posted the message below on my website a few weeks back, regarding a
company whom I used to work for. I was forced to back down and remove
it for legal reasons, but looking back I was stupid not to stand up and
fight this. However, even though the page was down, the cached version
remains. I was surprised to see that
this cached version had been removed in the last few days, especially
when some older pages (for which you get a "404") are still listed. Did
Google et al. bow to pressure from corporate bullies to remove the
link? It wouldn't surprise me! >:(
I'm posting this really because of a story that appeared a few days
ago, in which it is alledged on Slashdot that details of the UK
Police's "Operation Ore" (combating child pornography on-line) had been
censored by Google. I'm *hoping* that this isn't the case here.
"Life In the Autonomy Sweat Shop"
or: "Stress Is More Fun" by Dr.Paul Lee
Following a successful interview at Autonomy Headquarters in Cambridge
on March 24th, I was offered employment and agreed to start work on May
22nd. Despite this being a huge upheaval involving a huge outlay of
money (since no relocation fee was offered), I decided to make the move
from Woking to the Cambridge area.
At first, everything went well, and I was impressed by the company:
free lunches on Friday, TV during the 2006 England World Cup matches,
and even occasion social events (the Cambridge beer festival and
go-karting). But soon, the facade began to shatter and Autonomy
revealed itself as a company focussed on money and power. Visitors to
the company are insulated from this cut-throat attitude by the fun of
seeing red-bellied piranhas in the reception, and having board rooms
names after James Bond villains. The induction process consisted of
"Sign this contract, give us your bank details, heres your desk, here
are your co-workers, here is your staff handbook (about 12 pages), and
the sandwich van is up the road." About ten minutes in total: no
mention of quality, IT policy etc.
My first job was to devour the documentation concerning the company's
Virage product (a video archive and logging system) to write DLL
plug-ins to facilitate a company's video and audio analysis. This
involved reading through massive Software Development Kit documents,
and trying some of the sample plug-ins (bluescreen detection, for
instance). Everything was fine aoart from the so-called IT Support: it
took days to get my email sorted out, and weeks to get my Windows XP
workstation activated.
Sat next to me was Pieter, a Dutch (?) developer, whose voice was so
quiet I couldn't discern it over the noise of the fans and general din
in the office. Whenever I asked him for help, I passed him the keyboard
and asked him to type in the relevant stuff. He seemed to be doing work
on voice recognition, on the SoftCell system.
My manager was Abigail Betley, of whom more later.
My first job was write a plug-in that could detect and indentify logos,
or on-screen idents shown during a TV programme's broadcast. I quickly
identified many technical papers, and a simple method that would
isolate a non-animated logo from the rest of the screen. I tried out
this method and it worked fine. At this point, Abigail (Abby) went on
holiday for about 5 days or so, leaving me to doing some more work
(including coming up with Software Requirements, which to this day I
never received any feedback about). The only sore point was Abby's
colleague, Unai Ayo, a Spanish national, who phoned me up the day
before Abby was due back asking if "I had anything" and "whether logos
were correctly identified". Now, this was about a week and a half AFTER
I joined. Those of you who have done any research into Logo Detection
know that it is technically very complicated; indeed, one scientific
paper related how a Neural Net had been used (with about 89% accuracy)
- and here I was 10 days after joining the company and this major
technical problem was expected to be sorted out.
Alarm bells started ringing here.
Anyway, Abby came back and I was temporarily told to park the Logo
Detector (so that it could be handed over to the Neurodynamics team
downstairs, who had experience in image identification - car number
plates etc.). I was starting to get concerned as it looked like "my"
project was being taken off me. Besides, I had spoken to Neurodynamics
and it looked like they didn't know how to tackle the problem. Indeed,
Abby displayed many of the aspects that managers display - coming up
with ideas, but with no idea or solution as to how to implement them.
Another one of her ideas was a plug-in to detect and remove advert
breaks; a task that is proving hard to implement for the very best
software engineers and scientists on the planet. But, no matter.
My next project was, again, Virage related. The Crown Court service had
wanted some modifications done to the audio detection algorithm to
allow recording - and vocal recognition - of judges, defendants,
witnesses etc. I was expected to take over the work of Sergio Martinez,
another Spanish citizen. However, Sergio's accent was so strong, that I
could only figure out a few words at a time. He also wouldn't come to
my desk to help me as he said that my computer screen flickered badly
(!!!). Sergio had been shunted off to help with the blinkx web
browser/cataloging tool. Like most Autonomy tools, this uses the
Autonomy IDOL (Intelligent Data Operating Layer) mechanism at its core;
basically a fusion of Bayesian probabilities and Shannon's data
information theory to enable the computer to estimate some meaning
behind a document's contents and provide intelligent matches. (And to
be fair, its results are very impressive).
I was exasperated and indignant that I had been given someone else's
project. Sergio was in the office most days, and despite what was
required being simple in concept, I thought it would have been simple
to have Sergio spend a day doing the code, as he wrote it and was
familiar with it. Good idea, but management wanted me to do it. Abby
said that she wanted the modifications done "soon". In a meeting with
her, "soon", it transpired, meant "a week". Now at this point, I was
frustrated. The modifications involved testing the sound levels on left
and right channels, and the louder one would be logged to Virage's
servers. But I had no experience writing DLL's - indeed, I had only
written two, and one of those was the embryonic Logo Detector mentioned
above! My PC also lacked the drivers for its sound card - and even when
these were found, they were fiddly to install. I was given an audio
mixing deck. But without instructions, or cables. I had no idea how to
use it. Abby would help if she could, but even she was a bit perplexed
as to how Sergio's code worked. And due to Sergio's communication
problems, the best way to ask questions was by email (and he only sat
30 feet or so away!). Abby got a bit frustrated by my unfamiliarity
with DLLs and promised (and failed) to sit with me and go through the
code to help with the modifications. She promised to do this on two
other occasions, and failed both times. This suited me fine, as the way
she treated me with the whole mixing deck/DLL unfamiliarity fiasco
meant that I was losing respect for her.
Anyway, the modifications to the code were harder than I thought, and
for a good reason: as any good software engineer knows, code comments,
whilst not mandatory, are a good aid to either assist the memory when
going back to the code years later, or if a fresh eye wants to peruse
the algorithms. Apart from two comments (in Spanish!), there were NO
comments. This required time to backwards engineer the code. When I
asked Sergio about this, he admitted that he didn't had time to include
comments, that Autonomy didn't really subscribe to the notion of
comments, and that he only had TWO WEEKS to write his code! Makes you
wonder what kind of testing was performed. At least the lack of ISO/
TickIt accreditation started to make sense.
Anyway, that project ground to a halt, even though I did eventually do
the modifications, and commit the changes back to source control. Part
of the problems with testing lay in the fact that the audio files I was
provided with were unsuitable for testing. To determine which sound
levels in which channel - left or right - were the loudest, it would
have been nice to have well separated sound channels. That way, you can
determine which stereo channel should be picked up. The audio files
were useless for this. Either, there was a split stereo effect (so both
channels were the same) or only one channel was active. In the end, I
gave up and tried to use streaming audio from internet radio
broadcasts. This turned out to be very difficult and I gave up on
testing eventually. Unai's response to all this was "my results were
not good".
I had now been with Autonomy for a month and I was seriously thinking
of leaving. I had communicated my exasperation to Sarah Ambrose, of the
Paralegal team (and indeed, had asked what my notice period was). One
of the problems was Abby's desire to pester me up to three times a day
to ask if I had any results, making me feel that I was being victimised
and bullied. Sarah listened, and suggested that I speak to someone. She
mentioned the head of the technical group, Sean Blanchflower, but
(fortunately as it turned out) he was not available and I spoke to one
of his colleagues, who did not seem very fussed that code was not
commented and that I had been asked to modify someone's code when it
would have been far easier to ask the original author to do it.
After the Audio Analysis fiasco, I went back to the Logo Detector. I
wasn't given anything to do, so I resumed work on my own volition. I
eventually heard that we had a detector, developed by another company
Cobion. I was told it wasn't very good, but, because we had no
activation code, I couldn't see how it worked, or what it did. However,
the GUI did give me some ideas to what to do. For instance, allowing
the user to specify the area of the screen to analyse to speed up
processing. I developed my own dialog box, but this took a while
because drawing with MFC isn't very user friendly, using CDC etc.
Horrible. But I got it working after a few days, with a bit of help
from internet forums.
But still the hassling from Abby continued (I noticed she tended to
leave Pieter alone, but always pestered me).
My idea for Logo Detection involved the use of Phase Correlation (PC).
Using Logos isolated during the first week or two of my analysis, PC
would enable one to determine the amount of movement between two
similar objects in the same frame. For instance, if you had an isolated
logo, and did a PC with an image containing that logo, a peak should
appear in the correlation spectrum telling one how much displacement
there was between the two logos. My theory was, that by going through
each logo in turn, the one with the biggest spike in the correlation
spectrum would be the detected logo found in the streaming video.
Unfortunately, it didn't work out. I could prove that the concept
worked, but not with Virage. The PC method involves using Fast Fourier
Transforms (FFT). I tried three different ones: one from the internet,
which was the same as I found on the Autonomy library server; one from
"Numerical Recipes in C" and one cannibalised from the CImg project
(and in this, I agreed with Abby on one thing: due to CImg tool being
distributed under the GNU license, we couldn't use it, so I had to come
up with my own version). I tried these three FFTs and they all worked,
but wouldn't work in the plug-in. Was this a bug? I don't know. But if
it was, it wasn't the only one.
Let me explain. Although Virage would deal with colour images, it
wasn't quite so good with black and white. It couldn't output black and
white bitmaps (so I had to write my own algorithm), and the black and
white jpeg routine was faulty. Rather than show just one selected
grabbed frame in b&w, it would show three images, side by side of the
first couple of frames. Useless. I wanted black and white, as, because
Virage was a realtime application, I needed to speed up the algorithms
as fast as I could, so I reasoned that using black and white, rather
than colour images would be "faster". In the end, because of the lack
of suitable software, I used the Gimp tool to convert my reference
logos from colour to b&w as I couldn't trust Virage on this.
There were other problems with the internal software and documentation
too. I made a list of all flaws I found but, as will become clear, I
never got a chance to hand these over. I do remember one example: the
SDK manual described how to access the red, green and blue values for
each pixel. The only problem was that the manual listed the ordering of
the colour as being RGB. Now, I had very little experience with pixels,
bitmaps etc., so I had to teach myself. But it became obvious that the
ordering should be BGR. Anyone following Autonomy's manuals would get
the wrong result.
At this point, Summer was in full swing. Temperatures were in the late
20s F. And that was just in the office. I brought in a mercury
thermometer and found that, on nearly all the days, the temperature
would start at about 27. By early afternoon, it would usually reach 30.
I found it suprising that a company that made so much money and charged
a 6 figure sum for its software wouldn't invest in air conditioning,
just a few fans which succeeded in moving hot air around. I lost count
of the number of times I had to go to the toilet to cool off. I
frequently felt faint, and I was not the only one either.
Thankfully, there was a few days of relief from the bullying. I
attended a few days of a sales training course, which I enjoyed. The
presenter and trainer did tell us of Autonomy's competitors (eg "Fast",
which we had been told of constantly by email circulars, frequently got
small name companys for their search software while Autonomy got the
more famous ones - the BBC, BAE Systems, etc.). I did however, raise my
eyebrows when the trainer announced that he believed that, one day,
Autonomy would buy both Google and Microsoft! He also said that sales
staff would be in trouble if they consistently sold their software at
less than its expected press -an example figure of £80,000 was
mentioned.
When discussing competitor's products, the thrust was not on how good
Autonomy's software was - but how bad others was. Fast was mentioned a
lot and referred to in the sales literature: that it was slow to index,
that it needed huge memory upgrades to run, that it needed tweaks to
get it to work, that it only indexed a certain fraction of a page, and
only indexed enough pages to provide a "good enough result", and that
the company was in trouble for financial irregularities etc. These were
mentioned as things to say during a sales presentation.
Unfortunately, this training ate into my time on Logo Detector work.
The training went on from 9.30 (normal Autonomy start time) until
6.00pm (normal finish time). I ensured that I was in at 9.05 or
thereabouts and left after 6.00pm. However, on one of these days, Abby
got in early too. Without wishing me good morning, she said "Have you
got anything to show me?" I was truly disgusted by this, but nevermind.
I had lost every ounce of respect for her.
I had no idea what was to happen next. One Monday afternoon, Abby
called me into an office and lambasted for not producing any results
and that, "as an experienced software engineer", more had been expected
of me. I was disgusted by this and told her that I had no experience of
some of the software that we were using and had to learn some of it
from scratch. This didn't matter to her. I was so upset I emailed Sarah
Ambrose and, to my eternal regret, Sean Blanchflower, head of the
development group.
Two days later, Sean called me into an office, and basically said that
my accusations of bullying were "serious charges", but, as he said,
Abby was a valued member and produced results (translated: we're going
to back her). In my email to Sean, I had said that I was left feeling
despondent and very upset when I got home as I felt "where had it all
gone wrong?" I nearly burst into tears and he said that "if you don't
like the pressue you should consider leaving". There was no sympathy at
all, and he didn't really care about my concern: in fact, he reminded
me of my words at the interview where I had said that I was bored in my
previous job, which I felt was an excuse for Autonomy beating me into
the ground. As for Abby not greeting me that one morning and instead
asking me for results, he just put this down to the fact that some
people lack manners, an excuse that didn't ring true. The meeting
ended, but he hadn't listened to anything I had said.
About a week later, Alex Caunter, the head of Autonomy, summoned me
into the office. Sarah Ambrose had forwarded my emails to him. He asked
how things were going, and I said that they were going OK. Indeed, they
had. Abby had left me alone and I was free to get on with debugging and
testing. At this point, being on a 3 month probation period, I didn't
want to rock the boat so I said that things were going OK. He then
called Sean into his office and greeted him, saying that "you've been
here longer than I have" (which I took to mean that "I trust you") and
I casually mentioned how hot it was in the main Sweatshop. Sean
immediately countered by saying "We had three thermometers and none of
them are near 30": translated - "I'm not going to accept anything you
say". Despite his statement that my thermometer was broken, I feelt
that he would have disputed ANYTHING I said.
His first words were a bombshell "We're not very impressed by you". He
then reeled off a list of things that Abby had said, and some more -
that a software engineer straight from University could have done my
work faster than I did etc. He also said that I wasn't diligent as I
would often leave early. True I did leave early - at about 5.55pm, or 5
minutes early. But I explained that I needed to get my bus home. If I
got the next bus, I wouldn't get home until 8.40pm. Sean said that he
had viewed the company access logs, from using my keypass to the main
office (I said "So, you've been checking up on me?" and he said "of
course we have"- two points: firstly, this strikes me as inethical and
a violation of my rights, and secondly, I often left at the same time
as someone else and wouldn't use my keypass to open the door). Sean and
Alex didn't care on this point. I also mentioned that because of the
buses, I often got into work 15 or 25 minutes EARLY: I had even asked
Sarah Ambrose to adjust my pass to allow me to get into the office
early. Sean said "we wouldn't mind you leaving early if you get the
work done, but you don't". In fact, Autonomy expected to work unpaid
overtime it seems all the time. There then was the most spurious
srgument against me and which made me more sure than ever that I was
being forced out - "No one [in the office] knows who you are". That is
utterly irrelevant! However, one of Sean's other stupid, flithy lies
was easily countered- that I didn't talk to anyone: in fact, I did
confer with David, another Virage team member who would help with VS
Archive and loading data into it.
They did mention one salient thing: the reason why the pressure is so
high is to produce software before the quarterly results come out. SO,
everything is dominated by money. All during this meeting, Alex was
tapping away notes on his laptop, keeping quiet. The "hot and cold"
approach to staff, I thought.
Alex, at this point, mentioned that Abby should no longer be my manager
and that Sean should "monitor" me. He asked me to leave and I went back
to my desk nearly in tears.
The next day, I woke up feeling very low. I had been feeling upset for
weeks and had been having day dreams of jumping down the stairs and
putting a loaded shotgun under my chin. I went to my doctor's and he
signed my off work for two weeks under the diagnosis "stress". I also
had white patched on my hands - stress induced Vitiligo. I was also
given a prescription for diazepam to help me sleep. When I went back to
the doctors a few days later, he gave me another sick note, and I was
now on a huge doseage of Venlafaxine: 150mg per day.
Asking discussing the situation with my partner and my doctor, we
agreed that the culture at Autonomy was not good for me and I handed in
my resignation, by email to Sarah Ambrose on August 15th:
> Dear Sarah, > Please excuse this unorthodox method of communication, but I will not > be returning to work at Autonomy; my last day will the day my sick > note expires (August 29th). > > I simply cannot work in an environment that practises a blame culture, > and where employees are advised that "perhaps they should leave if > they don't like the pressure". Also having a management team that is > prepared to use exaggerations and irrelevancies is not too impressive > either. The overall effect has had a ruinous consequence on my health > > I truly regret this course of action, as, apart from the incidents > above, Autonomy has been a nice, relaxed environment. > > Yours SIncerely > > Paul Lee
My resignation was accepted, and a week later I got a letter from Alex
Caunter saying that my email contained unsubtantiated broadsides, and
that Autonomy were not happy with my productivity. They also said that
they wouldtake whatever steps were necessary to protect their
reputation (but in my eyes, you can't polish a turd).
So, the dream of Autonomy has turned sour. I am now in a house with an
expensive rent, with no job, and because I resigned, I will receive no
benefits. Fortunately I have savings to live on for a little while, but
I lack two other things: hope and prospects. In summary, I feel that I
have been chewed up and spat out by a company obsessed wtih money and
power. I should perhaps have listened to my instincts. In the summer of
2000, I attended an interview. When I got there (after travelling for 2
1/2 hours from New Malden), I was told that no-one knew who I was and I
was not expected. I waited for ages while the recruitment company was
contacted and the "interview" was back on. Even then, it was only a
cursory discussion. I didn't get the job. The company was Autonomy.
One last thing: you may wonder where "Stress is more fun" comes from.
In one of the manager's office, there is a notice on the desk, and even
though someone has inserted a "not" between "is" and "more", I believe
that the original meaning of the framed sign, visible from the main
office, is more applicable.
Postscript: a few weeks ago I posted this, I received a letter from
Autonomy's legal firm telling me to take down the article as it
divulged their working practises etc. and saying that damages would be
sought if I didn't comply. Yes, thats right: my little article has cost
the company millions in lost revenue. But I am horrified that a company
should take such bully-boy tactics given my precarious mental state. I
cannot help but wonder if the real reason was due to a private email
sent to me by Abby Betley: in it, she says that the article is causing
her some embarrasment. My heart bleeds.
