Tomorrow morning, I am keynoting at #hnycon with a "live" demo of how we @honeycombio bees release to production and debug when things inevitably break in the turbulent world that is customers & production traffic! Hope to see you at hnycon.io!
@honeycombio There is zero fakery involved in this demo, everything I did, I _actually_ did to our production systems, the only "cheat" is skipping forward by an hour mid-demo to skip past the "CI does its job, CD does phased deploy across environments with tests green" step.
pps before you passionately disagree saying "fast is not safe // releases are painful, so they need to happen as infrequently as possible", please do some background reading on DevOps and small batches, for instance from my friend @yesthattom :)
This isn't a brand new issue that Google can plead ignorance of or say it's scrambling to fix.
Trans Googlers as part of Gayglers, then as part of Trans@Google, raised these complaints for over a decade, including while I was head of the trans ERG at Google.
We were phenomenally lucky to have Demma Rosa as an ally on our side. But even with her escalating up through Ben Fried (CIO)'s org, we were unable to make traction on fixing the issue; it was almost as if Google just didn't care about being responsive to minority needs.
thank and coddle a senior white supremacist employee who claims the idea of white supremacy is "racist against white people", claim that it has to be like the Holocaust to be racist, then refuse to denounce white supremacy.
also, like, that is the kind of viewpoint that's being coddled as "we have to avoid making these people feel uncomfortable" -- it's not about what the marginal tax rate should be, they're literally treating overt white supremacist language as a form of difference you celebrate
and that's what's trendy among the exec set: saying "everyone has to be welcome no matter how they're different, especially my white supremacist friend" at the expense of the Black, brown, minority gender, and queer folks at their companies that they're driving out.
Don't dawdle, put something in before signing off!
Hi US East Coast, it's about to be 7pm, remember, we're not extending this deadline any further. Even if it's just some partial thoughts, put in for @o11ycon the topic of your talk and a few points you might want to cover! #OpenTelemetry#Observabilitygo.hny.co/3e5XcOY
Timestamps say I was stuck for 35 minutes on that two character mistake. and reviewing it, I see _exactly_ what happened. It wasn't just !(a || b) expansion, I was simultaneously refactoring !(f(x) == 2 || f(x) == 3) to only eval f(x) once, so I forgot the intention #AdventOfCode
Bought a 5600X from one of the more legit scalpers in .ca (AeroMalls, has a Newegg storefront, lists 20% above Canadian MSRP but not 50% or 100%+ above).
As the goblins say, "time is money, friend!" Swapping it in will wait for weekend, don't want to risk AoC stream or work day.
Sequence of operations: unplug both AMD64 machines, leaving the ARM64 workstation and the Chromebook as backup.
Swap B550 mobo into the larger case and the X470 mobo into the smaller case. Remove 3400G from X470, insert 4750G from B550. Plug 5600X into B550. Move RAM/GFX. Pray.
I wish I'd put the 4750G in the X470 to begin with (since both are PCIE 3.0 only), but they only recently released BIOS updates enabling 5000 and 4000 series on the X470 so that wasn't possible in September :/
even if the talk was added to pubapprove 2 days before the talk, if the approver got to it before the talk, the talk was approved. fullstop.
(if the approver _didn't_ get to it in the 2 days, then you'd have to retract/withdraw, of course. but not the pattern of facts here)
In fact, I am pretty damn sure that the vast majority of talks added to pubapprove _aren't_ submitted for review a full 2 weeks before deadline. I would eat my hat if every single Googler were diligent enough to work 2 weeks ahead of schedule.
"oh my god, there are videos of women being asphyxiated with plastic bags" -- have you considered for a moment that maybe some of those women _enjoy_ it and are getting paid to, and it's a fun day at work for them?
Seriously, if I didn't have a tech career, being a sex worker would also actually be pretty high up on my list, because I'd get paid to do something I like?
I seriously considered applying to work w/ kink\.com when I was in my first year of my tech career. It seemed fun.
The bar around revenue, customer base is held super high for founding teams made up of marginalized folks, but the instant you have white dudes with pedigrees leading your company, VCs throw in tens of millions of dollars without you having traction & product/market fit.
p.s. my q about unit economics has now been answered by the article: they charge you more the more you query, and pass on the costs of having an expensive indexing engine to you. Making customers think twice before querying is a hell of a business model.
til a "walk-in clinic" is different from a "primary clinic". why is health provision so difficult?
backstory is I know exactly what specialist center I need to go to to get my methotrexate refilled and get my bloodwork done, but it's gated on seeing a primary care doctor first to get a referral :/
that was _lucky_. This primary care doctor had treated autoimmune patients taking methotrexate before and was willing to write a prescription and order the correct labs (and also knew folks also take folic acid w/ MTX).
I will still need referral for eye exam, but whew.
#planeselfie because this is the only one I'm going to get for the next year at least...
Vancouver, here we come! Looking forward to being reunited with Elly after a month apart.
.@theflurbs looked very dejected when I abandoned her with CATSA at oversize baggage drop-off. Her second time on an airplane in her life, we think.
microcovid.org says at current rates in Ontario, with 3 other business class passengers near me within 6 feet for 6 hours, it's ~20 microcovid (10-70 confidence range), so it's between a few days and a month of error budget if I want 99.99% odds of surviving thru 2021.
whoa, I've healed enough that I momentarily blanked on my rapist's name.
It's Richard DeVaul (@rdevaul). After resigning from Google & deleting his Twitter after that story, he's back on tech twitter, and is the CTO of a company called Prometheus Fuels, according to his LinkedIn
@rdevaul Two years later, there's been no accountability. He's CTO of a Y Combinator backed startup, after having sexually assaulted me (and many others), & having sexually harassed @starsandrobots.
@daiwaka and @ktbenner's reporting has been met by a shrug by the "innovation ecosystem"
They're perfectly fine to hire a serial rapist and sexual harasser, who has literally had articles about his bad conduct printed in the NYT on page A1. Because their mission is about "innovation" and "solving climate change".
Hot take: if you're a SaaS provider of observability tooling, and didn't build your own data store & query engine specialized in the observability use case, your unit economics will suck.
(there may be other redeeming features of your product though, so it's not a fatal flaw)
the o11y use case is just... so different from every case on the market. It's not data warehousing, as you need results near-instantly rather than minutes or hours later. It's a write once, read specific rows almost never but when you do it matters, drilldown/slice ALL the time.
There's a reason that we didn't build Monarch or Dapper on top of Spanner or Bigtable in the end but instead had to go all the way down to RecordIO/ColumnIO at Google. Likewise if you look at every major player in our biz, they wrote their own telemetry store.