51
30

have-to-kill-this-guy

52
15
The Cult of Microsoft (www.wheresyoured.at)

Another great Ed Zitron essay about the tech industry. Some quotes:

The "growth mindset" is Microsoft's cult — a vaguely-defined, scientifically-questionable, abusively-wielded workplace culture monstrosity, peddled by a Chief Executive obsessed with framing himself as a messianic figure with divine knowledge of how businesses should work. Nadella even launched his own Bible — Hit Refresh — in 2017, which he claims has "recommendations presented as algorithms from a principled, deliberative leader searching for improvement."

There are many, many reasons this is problematic, but the biggest is that the growth mindset is directly used to judge your performance at Microsoft. Twice a year, Microsoft employees have a "Connect" with managers where they must answer a number of different questions about their current and future work at Microsoft, with sections titled things like "share how you applied a growth mindset," with prompts to "consider when you could have done something different," and how you might have applied what you learned to make a greater impact. Once filled-out, your manager responds with comments, and then the document is finalized and published internally, though it's unclear who is able to see them.

One employee related to me that managers occasionally add that they "did not display a growth mindset" after meetings, with little explanation as to what that meant or why it was said. Another said that "[the growth mindset] can be an excuse for anything, like people would complain about obvious engineering issues, that the code is shit and needs reworking, or that our tooling was terrible to work with, and the response would be to ‘apply Growth Mindset’ and continue churning out features."

In essence, the growth mindset means whatever it has to mean at any given time, as evidenced by internal training materials that that suggest that individual contributions are subordinate to "your contributions to the success of others," the kind of abusive management technique that exists to suppress worker wages and, for the most part, deprive them of credit or compensation.

One post from Blind, an anonymous social network where you're required to have a company email to post, noted in 2016 that "[the Growth Mindset] is a way for leadership to frame up shitty things that everybody hates in a way that encourages us to be happy and just shut the fuck up," with another adding it was "KoolAid of the month."

There are many, many reasons this is problematic, but the biggest is that the growth mindset is directly used to judge your performance at Microsoft. Twice a year, Microsoft employees have a "Connect" with managers where they must answer a number of different questions about their current and future work at Microsoft, with sections titled things like "share how you applied a growth mindset," with prompts to "consider when you could have done something different," and how you might have applied what you learned to make a greater impact. Once filled-out, your manager responds with comments, and then the document is finalized and published internally, though it's unclear who is able to see them.

The problem, it seems, is that Microsoft doesn't really care about the Growth Mindset at all, and is more concerned with stripping employees of their dignity and personality in favor of boosting their managers' goals. Some of Microsoft's "Connect" questions veer dangerously close to "attack therapy," where you are prompted to "share how you demonstrated a growth mindset by taking personal accountability for setbacks, asking for feedback, and applying learnings to have a greater impact."

This all feels so distinctly cult-y. Think about it. You have a High Prophet (Satya Nadella) with a holy book (Hit Refresh). You have an original sin (a fixed mindset) and a path to redemption (embracing the growth mindset). You have confessions. You have a statement of faith (or close enough) for new members to the church. You have a priestly class (managers) with the power to expel the insufficiently-devout (those with a sinful fixed mindset). Members of the cult are urged to apply its teachings to all facets of their working life, and to proselytize to outsiders.

As with any scripture, its textural meanings are open to interpretation, and can be read in ways that advantage or disadvantage a person.

And, like any cult, it encourages the person to internalize their failures and externalize their successes. If your team didn’t hit a deadline, it isn’t because you’re over-worked and under-resourced. You did something wrong. Maybe you didn’t collaborate enough. Perhaps your communication wasn’t up to scratch. Even if those things are true, or if it was some other external factor that you have no control over, you can’t make that argument because that would demonstrate a fixed mindset. And that would make you a sinner.

Yet there's another dirty little secret behind Microsoft's Connects.

Microsoft is actively training its employees to generate their responses to Connects using Copilot, its generative AI. When I say "actively training," I mean that there is an entire document — "Copilot for Microsoft 365 Performance and Development Guidance" — that explains, in detail, how an employee (or manager) can use Copilot to generate the responses for their Connects. While there are guidelines about how managers can't use Copilot to "infer impact" or "make an impact determination" for direct reports, they are allowed to "reference the role library and understand the expectations for a direct report based on their role profile."

To be extremely blunt: Microsoft is asking its employees to draft their performance reviews based on the outputs of generative AI models — the same ones underpinning ChatGPT — that are prone to hallucination.

Microsoft's culture isn't simply repugnant, it's actively dystopian and deeply abusive. Workers are evaluated based on their adherence to pseudo-science, their "achievements" — which may be written by generative AI — potentially evaluated by managers using generative AI. While they ostensibly do a "job" that they're "evaluated for" at Microsoft, their world is ultimately beholden to a series of essays about how well they are able to express their working lives through the lens of pseudoscience, and said expressions can be both generated by and read by machines.

I find this whole situation utterly disgusting. The Growth Mindset is a poorly-defined and unscientific concept that Microsoft has adopted as gospel, sold through Satya Nadella's book and reams of internal training material, and it's a disgraceful thing to build an entire company upon, let alone one as important as Microsoft.

53
21
submitted 2 weeks ago* (last edited 2 weeks ago) by AtmosphericRiversCuomo@hexbear.net to c/technology@hexbear.net

They fine-tuned a Llama 13B LLM with military specific data, and claim it works as well as GPT-4 for those tasks.

Not sure why they wouldn't use a more capable model like 405B though.

Something about this smells to me. Maybe a way to stimulate defense spending around AI?

54
66

Reddit just turned a profit for the first time. As part of its third quarter earnings results released on Tuesday, the company reported a profit of $29.9 million, along with $348.4 million in revenue — a 68 percent increase year over year.

The company hasn’t been profitable at any point in its nearly 20-year history. Since going public, Reddit lost $575 million during its first quarter on the market, but it decreased that loss to $10 million last quarter and is now finally in the black.

Reddit’s advertising revenue grew to $315.1 million, while “other” revenue reached $33.2 million on account of “data licensing agreements signed earlier this year.” Both Google and OpenAI have cut deals with Reddit to train their AI models on its posts.

55
29

I just think they're neat. I hate traditional mice and trackpads. The trackpoint nipple thingy they have got on thinkpads is my favorite pointing device but that is not available on most keyboards, but it should be. Trackballs are my second favorite

56
65
57
182
58
24
59
101

LoongArch: xigma-male

Intel: porky-scared-flipped

60
61
61
104
submitted 2 weeks ago* (last edited 2 weeks ago) by yogthos@lemmygrad.ml to c/technology@hexbear.net
62
26
63
113
64
142
submitted 2 weeks ago* (last edited 2 weeks ago) by plinky@hexbear.net to c/technology@hexbear.net
65
64
submitted 2 weeks ago by glans@hexbear.net to c/technology@hexbear.net

tldr: Same dev as pixelfed. It's not available yet, it won't be open source right away (but it's planned). Moderation TBD. iOS, android sideload. desktop later. Donations-based business model.

Loops, meanwhile, was developed by Daniel Supernault, who also created the federated Instagram rival Pixelfed. In fact, Loops will run under the Pixelfed project, according to an FAQ on its website.

In addition to the eventual benefit of being open source and distributed, Loops claims it will not sell or provide user data to third-party advertisers, use your content to train AI models, or gain the rights to any content uploaded on its service. Instead, users only grant Loops permission to use their content, but will retain full ownership of their contributions, the Loops website explains.

... Aimed at users 13 and up, Loops will allow you to follow other users, as well as like, comment on, or share their videos. But as a part of the federated web — the open social web running on ActivityPub — remote users from other platforms like Mastodon and Pixelfed will also be able to follow users’ Loops accounts and then view the videos in their home feed on those respective platforms. These remote followers will also be able to like, comment on, or share videos if their platform supports it.

Videos published to the app will be held for moderation if the uploader has a low trust score, but trusted users will be able to skip the queue and publish immediately. The trust score is also used to hide problematic comments on posts and apply content warnings, Supernault notes.

... According to Supernault, a side-loadable APK will be made available to Android users, and an iOS app will initially arrive on Apple’s TestFlight testing service when approved. A web interface won’t be an immediate focus but will come later on.

66
19
67
11

...versatile technique that combines a huge amount of heterogeneous data from many of sources into one system that can teach any robot a wide range of tasks

This method could be faster and less expensive than traditional techniques because it requires far fewer task-specific data. In addition, it outperformed training from scratch by more than 20 percent in simulation and real-world experiments.

Paper: https://arxiv.org/pdf/2409.20537

68
98

Those experts said some of the invented text — known in the industry as hallucinations — can include racial commentary, violent rhetoric and even imagined medical treatments.

69
54
submitted 2 weeks ago* (last edited 2 weeks ago) by yogthos@lemmygrad.ml to c/technology@hexbear.net
70
134
71
24
72
102
I hate agile (hexbear.net)
submitted 2 weeks ago by edge@hexbear.net to c/technology@hexbear.net

I hate scrum

I hate stand up

I hate sprints

Fuck Toyota

How have we taken the most autistic job and tacked a stupid, worthless, autistic unfriendly process on to it? (the answer of course is capitalism)

I want to quit but I can't get another job and even if I could it would just be more of this shit.

73
59
74
31
75
14

cross-posted from: https://hexbear.net/post/3770297

I think I've finally found it: The elusive Firefox fork for my day-to-day needs. It needed to have sane defaults like Librewolf but also as user empowering as Vivaldi (as well as not being proprietary which is cringe).

Zen I believe accomplishes both of that. It's a relatively new project but it does have active development with new changes added every release. Here's the rundown:

  • Licensed under the Mozilla Public License 2.0, the same as Firefox. So enjoy that warm feeling you get when using open source software that won't pull the rug from under you.
  • Follows Firefox release cycles: If a new Firefox version comes out, Zen is not behind.
  • Instead of horizontal tabs, Zen only uses vertical tabs for navigation. If this is a deal breaker, then Zen isn't for you :(
  • Supports split view, workspaces, browser profiles, side panels, tab unloading (saving memory by deactivating a tab), theming, mods and everything else that base Firefox supports (like firefox sync).
  • Cannot play DRM-protected content as of yet on Windows and MacOS (rare Linux W?) due to license fees. This is your netflix, your disney+, your spotify.
  • No mobile version (nor does it seem to be planned), though firefox sync is still supported.
  • Looks GORGEOUS. I never realized how ugly Firefox looks by default, esp on desktops like GNOME and KDE where it tries to integrate itself into the system theme.
  • Performs FABULOUSLY: Optimizations from the firefox level to even providing an optimized binary executable for modern CPUs.
  • SANE defaults like HTTPS everywhere, no link prefetching (where the browser loads links that it thinks you're going to go to), uncluttered Firefox home.
  • Probably more I'm not listing

Download here: https://zen-browser.app/download

How do I use Zen?

Well firstly, Zen doesn't come with any extensions by default. So I made sure to chuck in my Ublock Origin, Privacy Badger, ClearURLs, LibRedirect, etc. It also uses secure DNS by default with Cloudflare so you might want to turn that off (I have a DNS homeserver that does encrypted DNS through other means).

I also really like using the side panel to put my wiki sites and dictionaries in. I've only been using Zen for a week now and it seems to be my forever browser of choice.

view more: ‹ prev next ›

technology

23307 readers
295 users here now

On the road to fully automated luxury gay space communism.

Spreading Linux propaganda since 2020

Rules:

founded 4 years ago
MODERATORS